Smart city evaluation method and system based on AI analysis

Abstract

The invention relates to the technical field of AI analysis, and discloses a smart city evaluation method and system based on AI analysis, and the method comprises the steps: carrying out the intelligent analysis of city information, and generating the feature data of the current situation of a city; a group perception feedback mechanism is used as fitness, and optimized city feature data is generated through a long-acting adaptability optimization process; and comparing the optimized city feature data with the city current situation feature data, and analyzing the difference before and after optimization, thereby obtaining an evaluation result of the smart city. The city health condition is quantified by comparing the city current situation with the predicted ideal optimization state, and the improvement direction is defined and the resource allocation is optimized by identifying the key gap; and meanwhile, dynamic evaluation and trend prediction are supported. The city improvement potential is visually displayed, the trust and satisfaction of residents on city management are improved, and more efficient smart city management is realized.

Description

Technical Field

[0001] The present invention relates to the field of AI analysis technology, and specifically to a smart city evaluation method and system based on AI analysis. Background Art

[0002] With the rapid development of the Internet of Things (IoT), artificial intelligence (AI), big data and cloud computing technologies, the construction of smart cities is constantly advancing, aiming to use data analysis and prediction capabilities to optimize resource allocation, improve management efficiency, and ultimately achieve sustainable development and improve the quality of life of residents. In the existing smart city system, although there are many applications in the fields of transportation, environmental monitoring, energy management, etc., because each subsystem often operates independently, there are problems such as data islands, slow response and insufficient feedback mechanism. For example, the decentralized system architecture makes data integration difficult, resulting in incomplete decision-making information and affecting overall efficiency; in addition, traditional systems rely on static rules and are difficult to achieve real-time adaptive optimization; and the one-way feedback mechanism fails to fully collect and analyze residents' opinions and cannot dynamically respond to actual needs. The above problems hinder the further development and widespread application of smart city systems. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a smart city evaluation method based on AI analysis, comprising: Collect city information from the smart city evaluation platform; Performing intelligent analysis on the city information to generate city status characteristic data; Using the group perception feedback mechanism as fitness, through a long-term adaptive optimization process, optimized urban characteristic data is generated; During the optimization process, two types of feedback information are obtained through the group perception feedback mechanism, and the optimization direction is obtained through feature analysis. The abnormal parts are filtered out through the constraints of the smart city, and the remaining data after filtering are supplemented by the smart map of smart city construction. According to the selection mechanism, the set of smart nodes to be optimized is generated; Optimizing the set of intelligent nodes to be optimized according to the direction to be optimized; The optimized city characteristic data is compared with the city current characteristic data, and the differences before and after the optimization are analyzed to obtain the evaluation results of the smart city.

[0005] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, the city information includes numerical data, time series data, geographic space data, and image data.

[0006] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, the intelligent analysis includes establishing an algorithm library for AI intelligent analysis, entering each type of city information to match the corresponding algorithm, performing data processing on each type of city information using the matched algorithm, and converting the representation of each information object in the city information into a numerical form as city status characteristic data.

[0007] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, the group perception feedback mechanism includes obtaining the current perception and feedback of residents and tourists from multiple data sources, obtaining the group's description of the city's current status characteristic data through the current perception, and obtaining the group's suggestions on the current status through the feedback; Preprocessing the text data in the current perception and the feedback, identifying semantically repetitive content and removing it; Use named entity recognition algorithms to identify key objects in the text; In the text analysis, the main objects are quantitatively reflected, and the quantitative reflection results are selected and output in the set A; A={+1, -1, }, -1 indicates a quantized reflection in the negative direction, +1 indicates a quantized reflection in the positive direction, It means there is no quantitative reflection; Identify special limiting conditions in the feedback and determine the time period or area for optimization under these limiting conditions; Comparing the consistency of the content in the current perception and the feedback text, if the main object and the special limiting condition in the two types of text are the same, and the directions of the quantitative reflection results of the main object are opposite or there is no quantitative reflection, then the special limiting condition in the feedback is used as the optimization restriction of the main object, and the quantitative reflection result is used as the primary optimization direction of the main object; If the main objects and the special limiting conditions in the two types of texts are different, or the directions of the quantitative reflection results of the main objects are the same, the special limiting conditions in the feedback are used as the optimization restrictions of the main objects, and the quantitative reflection results are used as the secondary optimization direction of the main objects; the special limiting conditions in the current perception are used as the optimization restrictions of the main objects, and the opposite direction of the quantitative reflection results are used as the secondary optimization direction of the main objects; The quantitative reflection includes using dependency syntax parsing to identify the relationship between the quantity description word and the main object, and generating a quantitative reflection result of the quantity description word through a pre-trained neural network model.

[0008] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, wherein: the constraints of the smart city include establishing constraints on a feasible domain for the current status characteristic data of the city; In the process of the group perception feedback mechanism, the identification of the main object is constrained. If the main object is an adjustable urban factor, it is confirmed; if the main object is an unadjustable urban factor, the current identification result is not recorded; The text content parts that do not meet the constraints are treated as exceptions and filtered; The smart graph of the smart city construction includes taking all adjustable objects in the city information as smart nodes, connecting the smart nodes with synchronization relationships according to the preset relationship between each adjustable object, and embedding the positive and negative proportional relationship of the optimization direction at the connection point; The supplementing of the remaining data after filtering includes: processing only one optimization direction at a time, obtaining the main object corresponding to the optimization direction, marking the corresponding smart node in the smart map of the smart city construction; obtaining the marked smart node and other smart nodes connected thereto, and obtaining a smart node set for each optimization direction; after completing the integration of the smart node set marked for each optimization direction, updating the smart map to clear all marks, so that in each integration process, only one smart node has a mark; in the integration process of the smart node set, outside the marked smart node, other nodes are not integrated after the indirectly connected nodes; According to the positive and negative proportional relationship embedded in the connection, if the mark of the smart node is the primary optimization direction, the secondary optimization direction of other connected smart nodes is obtained; if the smart node is the secondary optimization direction, the tertiary optimization direction of other connected smart nodes is obtained; The selection mechanism includes grabbing and integrating all intelligent node sets one by one; wherein the optimization directions of different levels of each node can be superimposed; For node R, if node R is determined to be optimized L times by the group perception feedback mechanism, the smart node set marked with node R is captured L times; if node R is not marked in the smart node set, but is captured V times as an element in the smart node set; Then for node R, the number of grabs is equal to L+V; Assume that the weight of the first-level optimization direction is 3, the weight of the second-level optimization direction is 2, and the weight of the third-level optimization direction is 1; All crawling results are integrated into the same intelligent graph, and the optimization direction on each node is weighted and accumulated. The accumulated result is expressed as: ; Among them, i represents the crawled index; E represents the total number of crawls; j represents the superimposed optimization direction index; F represents the total number of superpositions of optimization directions of different levels for the currently crawled node; It represents the weight corresponding to the j-th optimization direction level superimposed on the j-th crawl; It represents the quantitative reflection of the j-th capture and the j-th optimization direction; If |Q|>G in the smart node, then the optimization is accepted, and the optimization direction is the same as the direction of Q; the smart nodes accepted for optimization are summarized to obtain the smart node set to be optimized.

[0009] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, wherein: for the smart node set to be optimized, the city status characteristic data of each node is optimized in the direction of Q for one step; The actual optimization situation is fed back through the smart city. For each smart node, if the actual optimization is completed for the smart node, the urban data of the smart city will be directly updated; if the actual optimization is not completed for the smart node and the optimization direction is consistent in the two previous optimization processes, then in the new round of optimization, the optimization step size of the smart node will be increased to two steps, and so on, until the actual optimization of the smart node is completed.

[0010] As a preferred solution of the smart city evaluation method based on AI analysis described in the present invention, wherein: comparing the optimized city characteristic data with the city status characteristic data includes obtaining all data before and after optimization; Current feature set: ; Optimized feature set: ; Calculate the distance D between the optimized feature and the current feature: ; Among them, n represents the number of data, k represents the data index, represents the nth status characteristic data, represents the k-th status characteristic data, represents the nth optimized city feature data, represents the kth optimized city feature data; Preset multiple intervals and set an evaluation level for each interval; compare D with the upper and lower limits of the set interval, analyze the interval in which the distance D between the optimized feature and the current feature is located, and output the evaluation level of the interval setting.

[0011] A smart city evaluation system based on AI analysis using any method described in the present invention, characterized in that: A collection unit collects city information from the smart city evaluation platform; performs intelligent analysis on the city information to generate city status characteristic data; The optimization unit uses the group perception feedback mechanism as fitness, generates optimized city characteristic data through a long-term adaptive optimization process, and optimizes the set of smart nodes to be optimized according to the direction to be optimized; The evaluation unit compares the optimized city characteristic data with the city current characteristic data, analyzes the difference before and after the optimization, and thus obtains the evaluation result of the smart city.

[0012] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein: the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0013] A computer-readable storage medium stores a computer program, wherein: when the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0014] Beneficial effects of the present invention: The AI-based smart city evaluation method provided by the present invention quantifies the health status of the city by comparing the current status of the city with the predicted ideal optimized state, and provides managers with a scientific basis for evaluation. By identifying key gaps, the present invention clarifies the direction of improvement and optimizes resource allocation; it also supports dynamic evaluation and trend prediction to help managers plan ahead. It intuitively demonstrates the potential for urban improvement, enhances residents' trust and satisfaction with urban management, and achieves more efficient smart city management. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0016] Figure 1 The overall process of a smart city evaluation method based on AI analysis is provided for the first embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0018] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a smart city evaluation method based on AI analysis, comprising: S1: Collect city information from the smart city evaluation platform.

[0019] Specific urban information includes: smart environment: air quality, water quality monitoring, green coverage, noise pollution, temperature and humidity, waste disposal; smart life: life satisfaction, distribution of medical resources, health index, distribution of educational resources, digital convenience services; smart economy: employment and unemployment rates, number of small and medium-sized enterprises, regional consumption activity, industrial structure, and financial service penetration rate; smart people: population density, education level, citizen participation, lifestyle and satisfaction; smart transportation: traffic flow and congestion, public transportation usage rate, travel mode data, parking space utilization rate, traffic accident rate, and number of electric vehicles; smart governance: government service data, emergency response efficiency, community service efficiency, and rule of law; smart tourism: number of tourists, distribution of tourism resources, tourist satisfaction, smart service level, cultural activity data, etc.

[0020] According to the type of data, it can be divided into numerical data, time series data, geospatial data, text data, and image data. Examples of numerical data include: air quality index (AQI), PM2.5 concentration, noise decibels, traffic flow (number of vehicles / hour); energy consumption (electricity, water, natural gas, etc.), such as consumption per unit time. Examples of time series data include: environmental monitoring data (such as hourly data of air quality, temperature, and humidity); traffic flow and congestion (traffic conditions by time period and region); daily, weekly, and monthly data on energy consumption; and trend data on the number of visitors to tourist attractions (by day or season). Geospatial data includes data on geographic location information, usually represented by latitude and longitude coordinates, and used for spatial analysis. Examples: geographic location of public facilities, attractions, hospitals, schools, etc.; regional green coverage data, distribution of pollution sources; road network structure, location of traffic congestion points; population density and building distribution in different regions. Text data is data stored in text form, which usually requires natural language processing methods for analysis. Examples: resident feedback, complaints, and suggestions; citizen satisfaction surveys on public services; policy announcements and news summaries. Image data is unstructured data containing image or video content, and is typically used for visual detection and analysis. Examples: city surveillance camera images and videos; satellite images (used to analyze green coverage, building density); traffic camera data (used to identify congestion); crowd flow monitoring images in public places.

[0021] S2: Perform intelligent analysis on the city information to generate city status characteristic data.

[0022] The intelligent analysis includes establishing an algorithm library for AI intelligent analysis, entering each type of city information to match the corresponding algorithm, performing data processing on each type of city information using the matched algorithm, and converting the representation of each information object in the city information into a numerical form as city status characteristic data.

[0023] Furthermore, the main purpose of this step is to generate characteristic data of the current status of the city through intelligent analysis of various types of urban information, so as to provide structured basic data support for subsequent optimization and evaluation. Different types of urban information (numerical, time series, geographic space, image, etc.) are converted into a unified numerical form through AI algorithms to form standardized characteristic data, which is convenient for system processing and analysis. With the development of AI algorithms, processing algorithms for different types of data have become mature and available. For example, convolutional neural networks (CNN) can process image data, long short-term memory networks (LSTM) are suitable for time series analysis, and GIS technology is suitable for feature extraction of geospatial data. Diverse algorithms can match and process complex data types. Modern AI technology can cope with large amounts of data through parallel computing, distributed processing, etc., especially in smart city systems. These technologies can generate accurate characteristic data while efficiently processing large amounts of real-time data.

[0024] Furthermore, the matching algorithm for numerical data in the algorithm library may be a data normalization and statistical analysis processing method. Normalization algorithms (such as Min-Max scaling or Z-score standardization) are applied to numerical data (such as air quality index, noise level, energy consumption, etc.) to standardize the data into a unified range for easy comparison and evaluation. Aggregate statistical analysis (such as mean value, standard deviation, etc.) is used to generate comprehensive characteristic values ​​of the data in different time periods or regions. For example, the average air quality index or energy efficiency of a city is generated.

[0025] The matching algorithm for time series data in the algorithm library can be a time series analysis (such as ARIMA, LSTM) processing method, which uses the ARIMA (autoregressive integrated moving average) model to perform trend analysis and prediction on time series data such as environment, transportation, and energy, extract periodic changes and future trend data, and use them for long-term evaluation of urban environment and resource management.

[0026] Using LSTM (Long Short-Term Memory Network) to predict complex time series is particularly suitable for irregular data (such as traffic flow fluctuations), and can generate numerical features such as traffic congestion trends and resource consumption warnings. The seasonality and trend characteristics in the time series analysis results are converted into indicators, such as traffic load index, environmental quality volatility, etc.

[0027] The matching algorithm of geospatial data in the algorithm library may be spatial clustering (DBSCAN) and geographic information system (GIS) analysis. The DBSCAN algorithm is used to perform spatial clustering analysis on geographic data, identify hot spots such as high-density pollution and congestion, and generate quantitative characteristics of urban subdivision environmental quality and traffic load. Spatial interpolation (such as Kriging interpolation) is performed through GIS technology to fill in the gaps in spatial data and generate a more complete pollution distribution map and green space coverage. The numerical data generated for each area represents the geographical characteristics of different blocks, such as environmental index, population density, etc.

[0028] Image data matching algorithms in the algorithm library: Convolutional neural network (CNN), target detection (such as YOLO), use CNN (convolutional neural network) to analyze satellite images, green images, etc., extract the visual features of the image, and convert it into numerical indicators such as green coverage and building density. Apply target detection algorithms (such as YOLO) to detect the number of vehicles and pedestrians in traffic monitoring images, and generate numerical features of traffic flow and pedestrian density. The extracted numerical representations include regional greening rate, building density, traffic flow, etc., which are used for quantitative evaluation of smart environment and smart transportation.

[0029] The feature data generated by the above analysis method will constitute the city status feature data set, and each feature data represents a certain aspect of the smart city status.

[0030] S3: Using the group perception feedback mechanism as fitness, generate optimized urban characteristic data through a long-term adaptive optimization process.

[0031] During the optimization process, two types of feedback information are obtained through the group perception feedback mechanism, and the optimization direction is obtained through feature analysis. The abnormal parts are filtered out through the constraints of the smart city, and the remaining data after filtering is supplemented by the smart map of smart city construction. According to the selection mechanism, a set of smart nodes to be optimized is generated.

[0032] Specifically, the group perception feedback mechanism includes obtaining the current perception and feedback of residents and tourists from multiple data sources, obtaining the group's description of the city's current status characteristic data through the current perception, and obtaining the group's suggestions on the current status through the feedback.

[0033] Preprocess the text data in the current perception and the feedback to identify and remove semantically repeated content. (Through cosine similarity calculation and DBSCAN clustering, semantically repeated feedback is automatically identified and repeated content is removed. This ensures the diversity of the data set and enables the smart city evaluation system to obtain richer and more independent feedback information.) Use named entity recognition algorithms to identify the main objects in the text; in an optional implementation, a pre-trained model such as BERT can be fine-tuned for named entity recognition. Specific implementation: Perform word segmentation and part-of-speech tagging on the feedback text.

[0034] Use the NER model to identify objects in the feedback (such as "bus", "transportation").

[0035] If the feedback does not explicitly mention the object, the keyword extraction model is used to determine the implicit main object.

[0036] Example: After processing the text "Buses are crowded in the morning", the main object "Bus" is obtained.

[0037] In the text analysis, the main objects are quantitatively reflected, and the quantitative reflection results are selected and output in the set A; A={+1, -1, }, -1 indicates a quantized reflection in the negative direction, +1 indicates a quantized reflection in the positive direction, Indicates no quantitative reflection.

[0038] What needs to be said is that the main object targeted by the text is identified, and the content orientation of the text is analyzed to be positive or negative (words such as "less" are negative, and words such as "more" and "increase" are positive), and special restrictions are analyzed. For example, "the bus is crowded in the morning", then the main object is the bus, the content orientation is negative (because the lack of buses will lead to this reason), and the special restriction is the morning time period. The above example involves the completion of the content "taking the bus in the morning, people are crowded", so as to analyze the adjustable urban factors (the number of buses), and analyze whether the object analysis is caused by positive reasons or negative factors. Through semantic analysis, there are too many people, but the large number of people is not an adjustable factor in the city, so the lack of cars is selected as the actual factor, so it becomes a negative direction. The quantitative reflection includes using dependency syntax parsing to identify the relationship between the quantity description word and the main object, and generating the quantitative reflection result of the quantity description word through the pre-trained neural network model. Through the AI ​​algorithm, this description can be directly identified and accustomed to from the training, so as to output the correct orientation.

[0039] Identify the special conditions in the feedback and determine the optimized time period or area under the conditions; for example, feedback: "During the morning rush hour, the subway station has too many people and the queue time is too long." Special conditions: morning rush hour. Optimized time period: morning rush hour (usually 7:00-9:00 on weekdays). Feedback: "Parking spaces in the city center are always full and it is difficult to park nearby." Special conditions: city center. Optimized area: parking facilities in the city center area.

[0040] Compare the consistency of the content in the text of the current perception and the feedback. If the main object and the special limiting conditions in the two types of text are the same, and the quantitative reflection results of the main object are in opposite directions or there is no quantitative reflection, then the special limiting conditions in the feedback are used as the optimization restrictions of the main object, and the quantitative reflection results are used as the primary optimization direction of the main object. In this case, the physical perception and the suggestion are two opposite directions, expressing the same appeal (for example, physical perception: crowded public transportation; suggestion: increase public transportation), which is more convincing with reason and evidence, so this type of data will get more weight.

[0041] If the main object and the special limiting conditions in the two types of texts are different, or the direction of the quantitative reflection results of the main object is the same, the special limiting conditions in the feedback are used as the optimization restrictions of the main object, and the quantitative reflection results are used as the secondary optimization direction of the main object; the special limiting conditions in the current perception are used as the optimization restrictions of the main object, and the opposite direction of the quantitative reflection results is used as the secondary optimization direction of the main object. In this case, a single narrative event or a claim can only reflect one-sided content and cannot prove authenticity and feasibility, but it is a reasonable claim, so it is based on a smaller weight.

[0042] Furthermore, the constraints of the smart city include establishing constraints on the feasible domain of the current characteristic data of the city. In the process of the group perception feedback mechanism, the identification of the main object is constrained. If the main object is an adjustable urban factor, it is confirmed; if the main object is an unadjustable urban factor, the current identification result is not recorded. The part of the text content that does not meet the constraints is treated as an exception and filtered. In the smart city system, some factors (such as the duration of traffic lights and bus schedules) can be adjusted by technical means, while other factors (such as weather and natural disasters) are uncontrollable. By screening and constraining objects in group feedback, only adjustable urban factors are retained as optimization objects, thereby ensuring that the system can produce actual effects instead of wasting resources on factors that cannot be adjusted. This constraint improves the accuracy of optimization and ensures that the optimization plan is executable and has practical significance.

[0043] The smart map of smart city construction includes taking all adjustable objects in the city information as smart nodes, connecting the smart nodes with synchronization relationships according to the preset relationship between each adjustable object, and embedding the positive and negative proportional relationship of the optimization direction at the connection point.

[0044] The supplementation of the remaining data after filtering includes processing only one optimization direction at a time, obtaining the main object corresponding to the optimization direction, and marking the corresponding smart node in the smart map of the smart city construction; obtaining the marked smart node and other smart nodes connected to it, and obtaining a set of smart nodes for each optimization direction; after completing the integration of the set of smart nodes marked for each optimization direction, updating the smart map to clear all marks, so that in each integration process, only one smart node has a mark (it can ensure that the core of the capture is the set of smart nodes constructed for the "this optimization direction". At the same time, it is of great significance for the weight of the subsequent statistical capture times, and different levels of optimization directions can be assigned in different capture processes. If the smart map is not updated, then each time it is captured, it will carry the mark of its own optimization direction level, which is not conducive to the operation of statistical rules); according to the positive and negative proportional relationship embedded in the connection, if the mark of the smart node is a primary optimization direction, the secondary optimization direction of other connected smart nodes is obtained; if the smart node is a secondary optimization direction, the tertiary optimization direction of other connected smart nodes is obtained. During the integration of the smart node set, in addition to the marked smart nodes, other nodes are not integrated after the indirectly connected nodes. This ensures that when grabbing a node, the longest connection length is 2 connection lengths, ensuring a high degree of relevance. By limiting the maximum connection length of the captured node to 2, it can be ensured that only smart nodes with a strong correlation with the target node are captured. This means that the system only processes those nodes that are highly correlated with the current optimization direction, avoiding invalid optimization or irrelevant optimization due to too large a capture range, saving system resources, and focusing on the optimization of key nodes.

[0045] It is important to know that different nodes in smart cities (such as transportation, energy, and environment) are usually interrelated, and changes in one node will have an impact on other nodes. By connecting smart nodes with synchronous relationships in the graph and embedding positive and negative proportional relationships at the connection points, the system can identify the chain reactions that may occur when optimizing one node on other nodes. For example, optimizing traffic flow may have a positive impact on environmental pollution and energy consumption. Through the identification of synchronous relationships, the system can coordinate and adjust related nodes while optimizing one node, thereby improving the coordination and overall efficiency of the entire system. In the smart graph, the optimization directions are divided into primary, secondary, and tertiary levels according to the optimization priority of the nodes, and the optimization requirements of each node can be deepened layer by layer. Through this hierarchical optimization strategy, the system can flexibly respond to optimization requirements at different levels. The primary optimization direction means that the optimization of the node needs to be completed first, while the secondary and tertiary optimization directions affect the related nodes in turn. This hierarchical optimization strategy enables the system to gradually refine the optimization steps, improve the sophistication and accuracy of the system, and ensure the priority optimization of important nodes.

[0046] The selection mechanism includes grabbing and integrating all the smart node sets obtained one by one (the captured sets can be the same sets obtained for different integration processes. Because each set reflects the demand for an optimization direction, the same set will not be deleted, which can also provide a large number of data items to support the subsequent accumulation operation); wherein the optimization directions of different levels of each node can be superimposed. For node R, if node R is determined to have an optimization direction L times by the group perception feedback mechanism, the smart node set marked by node R is captured L times; if node R is not marked in the smart node set, but is captured V times as an element in the smart node set. Then for node R, the number of captures is equal to L+V.

[0047] Assume that the weight of the first-level optimization direction is 3, the weight of the second-level optimization direction is 2, and the weight of the third-level optimization direction is 1. All crawling results are integrated in the same smart graph, and the optimization direction on each node is weighted and accumulated. The accumulated result is expressed as: ; Among them, i represents the crawled index; E represents the total number of crawls; j represents the superimposed optimization direction index; F represents the total number of superpositions of optimization directions of different levels for the currently crawled node; It represents the weight corresponding to the j-th optimization direction level superimposed on the j-th crawl; It represents the quantitative reflection of the j-th capture and the j-th optimization direction.

[0048] If |Q|>G in the smart node, then the optimization is accepted, and the optimization direction is the same as the direction of Q; the smart nodes accepted for optimization are summarized to obtain the smart node set to be optimized.

[0049] In the crawling results, the optimization direction of each node can be accumulated. This design helps to reflect the multiple optimization needs of the node in the smart map. If a node is frequently identified as an optimization node in different directions, the accumulated results will be weighted and superimposed to further improve the optimization priority of the node. Through this accumulation of optimization directions, key nodes can be avoided from being ignored, and continuous attention can be ensured for high-demand nodes, making their optimization more thorough. Moreover, the optimization directions of different levels of each node can be superimposed, which means that in different crawling times, it is possible to be captured due to the supplementary mechanism. For example, nodes a and b are connected. When a is crawled, it is captured as a primary optimization direction with a weight of 3. When b is crawled, a is captured at the same time as a secondary priority direction, so that a has two optimization directions superimposed, Q=3+2=5. This design can focus on projects that are mentioned many times and prevent them from being ignored.

[0050] S4: Optimizing the set of intelligent nodes to be optimized according to the direction to be optimized.

[0051] For the urban status characteristic data of each node in the set of smart nodes to be optimized, data optimization is performed in a step length in the direction of Q.

[0052] The actual optimization situation is fed back through the smart city. For each smart node, if the actual optimization is completed for the smart node, the urban data of the smart city will be directly updated; if the actual optimization is not completed for the smart node and the optimization direction is consistent in the two previous optimization processes, then in the new round of optimization, the optimization step size of the smart node will be increased to two steps, and so on, until the actual optimization of the smart node is completed.

[0053] If the optimization is achieved quickly, it means that the city's execution ability is very high and it can also prove the health of the city. On the contrary, if the execution ability is not high, it will lead to the accumulation of optimization, which means that the city's intelligence is still very low. Through the long-term accumulation and clearing mechanism, the self-renewal ability of the smart city can be reflected, which can serve as an important basis for evaluation. (Under such a design concept, if the city's self-renewal is fast, it will not cause the accumulation of optimization, and the expectations after optimization will be more realistic. However, if the city's self-renewal is very slow, the optimization will accumulate. At this time, the optimized expectations may not be able to accurately reflect the real expectations, but at this time, the comparison results of the optimized city characteristic data and the city's current characteristic data will be farther apart, thereby reducing the evaluation) S5: Compare the optimized city characteristic data with the city current characteristic data, analyze the differences before and after the optimization, and thus obtain the evaluation result of the smart city.

[0054] Comparing the optimized city characteristic data with the city current situation characteristic data includes obtaining all data before and after the optimization.

[0055] Current feature set: .

[0056] Optimized feature set: .

[0057] Calculate the distance D between the optimized feature and the current feature: ; Among them, n represents the number of data, k represents the data index, represents the nth status characteristic data, represents the k-th status characteristic data, represents the nth optimized city feature data, Represents the kth optimized city feature data.

[0058] Preset multiple intervals, and set an evaluation level for each interval; by comparing D with the upper and lower limits of the set intervals, analyze the interval in which the distance D between the optimized features and the current features is located, so as to output the evaluation level of the interval setting.

[0059] Suppose in the smart city system, the distance D between the optimized feature data and the current feature data is divided into the following three intervals, and each interval corresponds to an evaluation level: High health status: D ≤ 5. The gap between the current state and the ideal optimized state is very small, indicating that all aspects of the city are close to the healthy state and the overall situation is good.

[0060] Medium health status: 5 < D ≤ 15. There is a certain gap between the current state and the ideal optimized state. The health status of the city is尚可, but there are aspects that need to be improved.

[0061] Poor health status: D > 15. There is a large gap between the current state and the ideal optimized state, indicating that the health status of the city is not ideal, there are significant problems, and key attention and improvement are required.

[0062] Suppose in the health status evaluation of the smart city, D = 12 is obtained through system calculation. According to the preset interval, the health level of this state can be determined as "medium health status". This evaluation result indicates that although the current indicators of the city are not ideal, they basically meet the health standards and there is still a certain room for improvement.

[0063] On the other hand, this embodiment also provides a smart city evaluation system based on AI analysis, which includes: A collection unit that collects city information in the smart city evaluation platform; performs intelligent analysis on the city information to generate current city feature data.

[0064] An optimization unit that uses the group perception feedback mechanism as the fitness, and through a long-term adaptive optimization process, generates optimized city feature data; optimizes the set of smart nodes to be optimized according to the direction to be optimized.

[0065] An evaluation unit that compares the optimized city feature data with the current city feature data, analyzes the differences before and after optimization, so as to obtain the evaluation result of the smart city.

[0066] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0067] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0068] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0069] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0070] Embodiment 2 is an embodiment of the present invention, which provides a method for evaluating a smart city based on AI analysis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0071] During the experiment, real-time data of the city under multiple indicators was first collected, and these data cover the operating conditions of all aspects in the smart city. Taking the evaluation system of the present invention as the core, data standardization and cleaning were first carried out to remove abnormal data points and ensure the validity and consistency of the experimental data. Then, these data were processed in layers to calculate the current values of various features. At the same time, combined with the AI analysis model embedded in the system, an ideal optimized target state was generated based on the current situation data for evaluating the health level of the current situation.

[0072] After the data collection is completed, the system calculates the difference between the current state feature data and the optimized target feature data according to the grading evaluation weights in four aspects: transportation, environment, resources, and social services, and converts this difference into a health evaluation distance D. During the analysis, the preset health evaluation levels of the system are used as measurement indicators, and combined with the mutual influence between features, a health evaluation level matrix is formed. Specifically, the health evaluation is divided into three levels: "high", "medium", and "poor", corresponding to D≤5, 5<D≤15, and D>15 respectively, to quantify the optimization potential of each feature.

[0073] In the data comparison and analysis stage, through the intelligent graph of the present invention, the system can automatically identify the synchronization relationship between features (such as the influence relationship between traffic flow and air quality) and dynamically adjust the health weights of different nodes accordingly. Finally, the experimental results are exported as a health evaluation report, which includes health level assessment, key optimization suggestions, and future trend prediction, so as to provide intuitive, data-driven health status assessment and decision support for urban managers. The process data is recorded in Table 1.

[0074] Table 1 Data Record Table

[0075] The data comparison in Table 1 clearly shows the effectiveness of the present invention in the evaluation of health status of smart cities. In the table, the current characteristic value and the target optimization value represent the current status and ideal health status of each indicator. By calculating the difference between the current status and the target value and weighted accumulation, the system finally obtains the comprehensive health score of each indicator and divides the health level into "high", "medium" and "poor".

[0076] Through multiple simulation optimizations and evaluation of the changes in urban health evaluation before and after optimization, the improvement effect of the evaluation system of the present invention in long-term optimization is observed. The experimental period is set to 6 months, divided into 6 stages. In each stage, the urban evaluation data is recorded and analyzed, and the optimization direction is adjusted according to the before-after comparison results.

[0077] During the experiment, the system collected key characteristic data of the city once a month, covering key indicators such as traffic flow, air quality, energy consumption, utilization of education and medical resources, garbage disposal volume, and water resource utilization. According to the ideal state predicted by the system, the phased optimization direction is set, hoping to gradually approach a healthy state through each optimization.

[0078] At each stage, the system first analyzes the difference between the current status data and the optimization target state, and calculates the health evaluation level according to the intelligent evaluation algorithm of the present invention. If the health evaluation level of a certain indicator remains unchanged for two consecutive times, the system will automatically increase the optimization intensity of the indicator, adjust the corresponding optimization weight, and generate a new round of optimization direction. After each cycle, the system re-evaluates the health of each indicator, records the data before and after optimization for horizontal comparison, and the recorded data is shown in Table 2.

[0079] Table 2 Optimization process record stage Traffic flow (vehicles / hour) Air Quality Index (AQI) Energy consumption (MWh) Water resource utilization (10,000 cubic meters) Medical resource utilization rate (%) Utilization rate of educational resources (%) Garbage disposal volume (tons) Comprehensive health score Initial status 8500 90 320 18 75 80 500 Difference Before Phase 1 Optimization 8200 85 315 17.5 74 79 490 Difference After Phase 1 Optimization 8000 80 310 17 73 78 480 middle Before Phase 2 Optimization 8000 80 310 17 73 78 480 middle After Phase 2 Optimization 7800 75 305 16.8 73 78 470 middle Before Phase 3 Optimization 7800 75 305 16.8 73 78 470 middle After Phase 3 Optimization 7500 70 300 16.5 73 78 460 middle Before Phase 4 Optimization 7500 70 300 16.5 73 78 460 middle After Phase 4 Optimization 7300 65 300 16.5 72 77 450 high Before Phase 5 Optimization 7300 65 300 16.5 72 77 450 high After Phase 5 Optimization 7200 60 300 16.5 72 76 440 high Before stage 6 optimization 7200 60 300 16.5 72 76 440 high After stage 6 optimization 7100 55 300 16.5 72 76 430 high As can be seen from Table 2, after 6 stages of optimization, the smart city health status evaluation system of the present invention effectively improves multiple key city indicators and achieves a gradually improved health level.

[0080] The health evaluation of the city was "poor" in the initial state, and it was improved to "medium" after the optimization in the first and second stages. After the fourth stage, the health level reached "high" and remained unchanged in the following stages. Through multiple optimizations, the system has achieved significant improvements in multiple key indicators, making the health evaluation gradually approach the ideal state, verifying the continuous role of the present invention in long-term optimization. After the third stage, energy consumption (300 MWh) and water resource utilization (165,000 cubic meters) have approached the target optimization value, so the system will no longer further optimize these indicators from the fourth stage, and resources will be transferred to other indicators that need to be improved. This strategy not only avoids resource waste, but also ensures the balance between various indicators, so that the overall health status is further improved. It can be observed from the table that the amount of garbage processed gradually decreases in the fifth and sixth stages, but other indicators that are close to the ideal state remain unchanged. This staged change makes the health state of the system more stable after optimization and avoids repeated fluctuations caused by excessive adjustments. This verifies the stability of the present invention in long-term health management.

[0081] The smart city health evaluation system of the present invention has shown significant improvement effects and continuous health management capabilities in multiple rounds of optimization. Compared with the existing technology, the present invention not only has an automatic control and feedback mechanism, but also ensures the reasonable maintenance of key indicators during the gradual optimization process, avoids resource waste, and ultimately achieves comprehensive health evaluation and dynamic optimization of smart cities, providing managers with more scientific and operational decision support.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. 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 smart city evaluation method based on AI analysis, characterized in that: include: Collect city information from the smart city evaluation platform; Performing intelligent analysis on the city information to generate city status characteristic data; Using the group perception feedback mechanism as fitness, through a long-term adaptive optimization process, optimized urban characteristic data is generated; During the optimization process, two types of feedback information are obtained through the group perception feedback mechanism, and the optimization direction is obtained through feature analysis. The abnormal parts are filtered out through the constraints of the smart city, and the remaining data after filtering are supplemented by the smart map of smart city construction. According to the selection mechanism, the set of smart nodes to be optimized is generated; Optimizing the set of intelligent nodes to be optimized according to the direction to be optimized; Comparing the optimized city characteristic data with the current city characteristic data, analyzing the differences before and after the optimization, and thus obtaining the evaluation results of the smart city; The group perception feedback mechanism includes obtaining the current perception and feedback of residents and tourists from multiple data sources, obtaining the group's description of the city's current status characteristic data through the current perception, and obtaining the group's suggestions on the current status through the feedback; Preprocessing the text data in the current perception and the feedback, identifying semantically repetitive content and removing it; Use named entity recognition algorithms to identify key objects in the text; In the text analysis, the main objects are quantitatively reflected, and the quantitative reflection results are selected and output in the set A; A={+1, -1, }, -1 indicates a quantized reflection in the negative direction, +1 indicates a quantized reflection in the positive direction, Indicates no quantitative reflection.

2. The smart city evaluation method based on AI analysis as claimed in claim 1, characterized in that: The city information includes numerical data, time series data, geographic space data, and image data.

3. The smart city evaluation method based on AI analysis as claimed in claim 2, characterized in that: The intelligent analysis includes establishing an algorithm library for AI intelligent analysis, entering each type of city information to match the corresponding algorithm, performing data processing on each type of city information using the matched algorithm, and converting the representation of each information object in the city information into a numerical form as city status characteristic data.

4. The smart city evaluation method based on AI analysis as claimed in claim 3, characterized in that: The group perception feedback mechanism also includes identifying special limiting conditions in the feedback and determining the optimized time period or area under the limiting conditions; Comparing the consistency of the content in the current perception and the feedback text, if the main object and the special limiting condition in the two types of text are the same, and the directions of the quantitative reflection results of the main object are opposite or there is no quantitative reflection, then the special limiting condition in the feedback is used as the optimization restriction of the main object, and the quantitative reflection result is used as the primary optimization direction of the main object; If the main objects and the special limiting conditions in the two types of texts are different, or the directions of the quantitative reflection results of the main objects are the same, the special limiting conditions in the feedback are used as the optimization restrictions of the main objects, and the quantitative reflection results are used as the secondary optimization direction of the main objects; the special limiting conditions in the current perception are used as the optimization restrictions of the main objects, and the opposite direction of the quantitative reflection results are used as the secondary optimization direction of the main objects; The quantitative reflection includes using dependency syntax parsing to identify the relationship between the quantity description word and the main object, and generating a quantitative reflection result of the quantity description word through a pre-trained neural network model.

5. The smart city evaluation method based on AI analysis as claimed in claim 4, characterized in that: The constraints of the smart city include constraints on the feasible domain of the current characteristic data of the city; In the process of the group perception feedback mechanism, the identification of the main object is constrained. If the main object is an adjustable urban factor, it is confirmed; if the main object is an unadjustable urban factor, the current identification result is not recorded; The text content parts that do not meet the constraints are treated as exceptions and filtered; The smart graph of the smart city construction includes taking all adjustable objects in the city information as smart nodes, connecting the smart nodes with synchronization relationships according to the preset relationship between each adjustable object, and embedding the positive and negative proportional relationship of the optimization direction at the connection point; The supplementing of the remaining filtered data includes processing only one optimization direction at a time, obtaining the main object corresponding to the optimization direction, marking the corresponding smart node in the smart map of the smart city construction; obtaining the marked smart node and other smart nodes connected thereto, and obtaining a set of smart nodes for each optimization direction; After completing the integration of the smart node set marked in each optimization direction, update the smart graph to clear all marks, so that in each integration process, only one smart node has a mark; in the integration process of the smart node set, outside the marked smart nodes, other nodes are not integrated except for the indirectly connected nodes; According to the positive and negative proportional relationship embedded in the connection, if the mark of the smart node is the primary optimization direction, the secondary optimization direction of other connected smart nodes is obtained; if the smart node is the secondary optimization direction, the tertiary optimization direction of other connected smart nodes is obtained; The selection mechanism includes grabbing and integrating all intelligent node sets one by one; wherein the optimization directions of different levels of each node can be superimposed; For node R, if node R is determined to be optimized L times by the group perception feedback mechanism, the smart node set marked with node R is captured L times; if node R is not marked in the smart node set, but is captured V times as an element in the smart node set; Then for node R, the number of grabs is equal to L+V; Assume that the weight of the first-level optimization direction is 3, the weight of the second-level optimization direction is 2, and the weight of the third-level optimization direction is 1; All crawling results are integrated into the same intelligent graph, and the optimization direction on each node is weighted and accumulated. The accumulated result is expressed as: ； Among them, i represents the crawled index; E represents the total number of crawls; j represents the superimposed optimization direction index; F represents the total number of superpositions of optimization directions of different levels for the currently crawled node; It represents the weight corresponding to the j-th optimization direction level superimposed on the j-th crawl; It represents the quantitative reflection of the j-th capture and the j-th optimization direction; If |Q|>G in the smart node, then the optimization is accepted, and the optimization direction is the same as the direction of Q; the smart nodes accepted for optimization are summarized to obtain the smart node set to be optimized.

6. The smart city evaluation method based on AI analysis as claimed in claim 5, characterized in that: For the urban status characteristic data of each node in the set of smart nodes to be optimized, data optimization is performed in the direction of Q for one step; The actual optimization situation is fed back through the smart city. For each smart node, if the actual optimization is completed for the smart node, the urban data of the smart city will be directly updated; If the actual optimization of the smart node is not completed and the optimization directions are consistent in the two previous optimization processes, the optimization step size of the smart node is increased to two steps in the new round of optimization, and so on, until the actual optimization of the smart node is completed.

7. The smart city evaluation method based on AI analysis as claimed in claim 6, characterized in that: Comparing the optimized city characteristic data with the city current characteristic data includes obtaining all data before and after the optimization; Current feature set: ; Optimized feature set: ; Calculate the distance D between the optimized feature and the current feature: ； Among them, n represents the number of data, k represents the data index, represents the nth status characteristic data, represents the k-th status characteristic data, represents the nth optimized city feature data, Represents the kth optimized city feature data; Preset multiple intervals and set an evaluation level for each interval; compare D with the upper and lower limits of the set interval, analyze the interval in which the distance D between the optimized feature and the current feature is located, and output the evaluation level of the interval setting.

8. A smart city evaluation system based on AI analysis using the method according to any one of claims 1 to 7, characterized in that: A collection unit collects city information from the smart city evaluation platform; performs intelligent analysis on the city information to generate city status characteristic data; The optimization unit uses the group perception feedback mechanism as fitness and generates optimized urban characteristic data through a long-term adaptive optimization process; Optimizing the set of intelligent nodes to be optimized according to the direction to be optimized; The evaluation unit compares the optimized city characteristic data with the city current characteristic data, analyzes the difference before and after the optimization, and thus obtains the evaluation result of the smart city.

Images