A method for constructing and optimizing the operation of digital twin cities based on cloud computing and big data technology

Abstract

The present invention provides a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology, the method comprising: optimizing the layout of various sensor devices and adjusting the acquisition frequency through a particle swarm optimization algorithm and an adaptive sampling algorithm; building a distributed database, and achieving uniform distribution and efficient access of data in the distributed database through a load balancing algorithm; using a time series model to mine the temporal laws of urban operation and discover the association between different data; constructing a city model by using a three-dimensional real-scene fusion modeling method based on oblique photography of unmanned aerial vehicles and supplemented by ground laser radar, applying a machine learning method to identify different areas in the city, and constructing refined models of various fields of the city; realizing real-time simulation of the operation status of the city through virtual reality and augmented reality technology; and continuously optimizing the system using an online learning algorithm so that it can continuously learn and adapt to new data patterns during operation, thereby improving its construction and operation optimization capabilities.

Description

Technical Field

[0001] The present invention relates to the fields of cloud computing, big data technology and digital twin city systems, and specifically to a digital twin city construction and operation optimization method based on cloud computing and big data technology, which is used to construct and optimize a digital twin city system. Background Art

[0002] The city digital twin system has greatly improved the efficiency and effectiveness of urban management through its excellent data visualization capabilities. It not only provides an intuitive and dynamic management tool for city managers, but also helps decision makers gain insight into the future trends of urban development through simulation and prediction, thereby achieving more scientific and reasonable urban planning and resource allocation. The improvement and development of the city digital twin system has a far-reaching impact on promoting the construction of smart cities, improving the level of urban governance, and enhancing the quality of life of urban residents.

[0003] However, with the increase in application scenarios and the increase in complexity, urban digital twin systems are facing endless practical challenges. First, the data collected by IoT devices are complex and easily affected by many uncertain factors, which undoubtedly brings great difficulties to data monitoring and analysis; secondly, the complexity and diversity of data make traditional databases incapable of parsing the subtle relationship between data and achieving efficient storage; and the data collected by IoT sensors are of various types, and the relationship between data is difficult to measure accurately; in addition, in the process of building a three-dimensional model of the city, traditional aerial photography technology can no longer meet the requirements for precision, and it is difficult to accurately classify different urban areas; finally, although the traditional digital twin system has the function of data visualization, it lacks the deep integration of virtual reality and augmented reality technology. More importantly, these systems are often limited to the display of existing data and lack the ability to continuously optimize and self-evolve. Summary of the invention

[0004] The purpose of the present invention is to provide a method for constructing and operating a digital twin city based on cloud computing and big data technology. The method uses efficient particle swarm optimization and adaptive sampling algorithms to construct a collection network and a cloud computing platform. The intrinsic connections between the data collected by the equipment are then mined, and a three-dimensional model of the city is established. Real-time simulation is achieved by combining virtual reality and other technologies. Finally, based on an online learning algorithm, the system is continuously optimized, which can significantly reduce the errors and costs of manual inspections, and at the same time provide strong technical support for real-time query of urban status and urban safety management.

[0005] The object of the present invention is achieved through the following technical solutions:

[0006] The digital twin city construction and operation optimization method based on cloud computing and big data technology includes the following steps:

[0007] Step S1: Build a data collection network, optimize the layout of various sensors and IoT devices based on the particle swarm optimization algorithm and adaptive sampling algorithm, adjust the sampling frequency, and collect urban operation data in real time;

[0008] Step S2: Using the cloud computing platform to build a distributed database, using a load balancing algorithm to evenly distribute data in the distributed database, and achieving efficient data storage and management;

[0009] Step S3: pre-process the collected data based on big data analysis technology, use the time series model to mine the time law of urban operation and use the intrinsic association mining algorithm to find the association between different data, so as to provide a basis for model construction;

[0010] Step S4: construct a city model by using a 3D real-scene fusion modeling method based on drone oblique photography and supplemented by ground-based laser radar, apply machine learning methods to identify different communities or commercial areas in the city, and construct refined models of various areas in the city;

[0011] Step S5: Realize real-time simulation of city operation status through virtual reality and augmented reality technology;

[0012] Step S6: Use online learning algorithms to continuously optimize the system so that the system can continuously learn and adapt to new data patterns during operation, thereby improving the construction and operation optimization capabilities of the digital twin city.

[0013] Further, the step S1 specifically includes:

[0014] Step S101: Establish the deployment goals of sensors and IoT devices and define the objective function Used to evaluate the performance of layout solutions, where The parameter vector representing the equipment layout, the objective function calculation formula is:

[0015] ;

[0016] in, is the coverage rate, is the device density, is the acquisition frequency, , , is the corresponding weight coefficient;

[0017] Step S102: Optimizing the layout of sensors and IoT devices using a particle swarm optimization algorithm. The specific steps of optimizing the layout using the particle swarm optimization algorithm are as follows:

[0018] (1) Initialize a group of particles, each particle represents a device layout scheme, where i=1, 2, ..., ;

[0019] (2) Evaluate the fitness of each particle , i.e., the performance of the layout scheme;

[0020] (3) Update particle speed and equipment layout plan , the updated calculation formula is:

[0021] ;

[0022] ;

[0023] in, for The speed of the time particle, for Time equipment layout plan, for Time equipment layout plan, is the inertia weight, and is the learning factor, and is a random number between [0, 1], is the optimal position of an individual, For the optimal layout solution, is the disturbance coefficient, For the The dynamic disturbance vector of a particle, is the position update coefficient; for The speed of time particles;

[0024] (4) Repeat the above steps until the termination condition is met. = , is the termination condition time, and the optimal layout solution is obtained ;

[0025] Step S103: Using the optimal layout solution , an adaptive sampling algorithm is used to adjust the acquisition frequency of the device.

[0026] Furthermore, in step S103, the step of using the adaptive sampling algorithm to adjust the acquisition frequency of the device is:

[0027] First, according to real-time data changes and preset thresholds First, according to the real-time data changes and preset thresholds , adaptive adjustment Time Single Device Collection Frequency , the adaptive adjustment formula is:

[0028] ;

[0029] in, for Time single device collection frequency, is the adjustment factor, To adjust the index, is a decay term that increases with time, where is the attenuation coefficient, is the time step, the decay term will increase with the increase of time t, thus reducing The impact of

[0030] Among them, the acquisition frequency adjustment factor of each device The calculation formula is:

[0031] ;

[0032] in, For the The optimal acquisition frequency for each device, For the The current acquisition frequency of each device;

[0033] Then, design Objective function, find its minimum value, used for evaluation performance;

[0034] ;

[0035] in, is the coverage rate, which indicates the proportion of urban areas that can be covered by the acquisition frequency; The device load represents the workload of the device under the set acquisition frequency. The value range of this workload is determined by the physical device and does not exceed the maximum workload. , , is the corresponding weight coefficient, which is used to balance the importance of different objectives; Data value, which indicates the importance of the collected data to urban management and decision-making;

[0036] It is expressed as:

[0037] ;

[0038] in, , , They represent the impact of data on decision-making, the real-time nature of data, and the integrity of data, respectively, and their value range is [0, 1]. , , Represent their weight coefficients respectively;

[0039] Finally, real-time collection of urban operation data ensures the real-time and accuracy of the data and provides basic data support for subsequent analysis and simulation. The data collection process is expressed as:

[0040] ;

[0041] ;

[0042] in, For the Devices at time The collected data, For single Devices at time The collected data, For time The acquisition frequency, For time A collection of data collected from different devices.

[0043] Further, the step S2 specifically includes:

[0044] Step S201: Design a distributed database architecture. The specific steps are as follows:

[0045] First, determine the node topology and assign a unique identifier. The calculation formula for the topology and identifier is:

[0046] ;

[0047] ;

[0048] in, is the node topology, For the nodes, is the total number of nodes, For Node A unique identifier for Compute a function for an identifier;

[0049] Then, the node hash value is calculated, where the calculation formula of the node hash value is:

[0050] ;

[0051] in, For Node The hash value of is a hash function, is the number of bits of the hash ring;

[0052] Step S202: Implement the consistent hashing algorithm. The specific steps are as follows:

[0053] First, initialize the hash ring, where the initialization formula is:

[0054] ;

[0055] in, is a hash ring;

[0056] Then, map the node to the hash ring and determine the node position, where the mapping calculation formula is:

[0057] ;

[0058] ;

[0059] in, For Node Position on the hash ring;

[0060] Step S203: Data distribution and storage, mapping data items to hash rings and determining data storage locations, the formula is as follows:

[0061] ;

[0062] ;

[0063] in, For data items The position on the hash ring, For data items The hash value of To store data items Nodes; To find the node function;

[0064] Step S204: Load balancing and data migration. The specific steps of the operation are:

[0065] First, the data migration when adding a new node is calculated as follows:

[0066] ;

[0067] in, Indicates the data migration of newly added nodes. is the hash value difference between the newly added node and the node where the data item is currently located, For data items The amount of migration;

[0068] Then, delete the data migration of the node. The calculation formula for deleting data migration is:

[0069] ;

[0070] in, The hash value difference between the deleted node and the node where the data item is currently located;

[0071] Step S205: Data access and query optimization. The specific steps of the operation are:

[0072] First, build an index of device collection data for query optimization. The method is as follows:

[0073] ;

[0074] ;

[0075] in, For data items The index of Build a function for the index, Used to calculate query time, Used to obtain the index size;

[0076] Then, the cache strategy for collected data is:

[0077] ;

[0078] ;

[0079] in, For data items The cache status, is the time step, is the cache strategy function, for The number of cache accesses, for Total visits, Indicates the ratio of the number of successful data acquisitions from the cache to the total number of data acquisition requests within a set time, which is an indicator used to measure cache performance;

[0080] Finally, the query routing process of the device collection data, the query expression is:

[0081] ;

[0082] in, To query the routing function, A function to find the nearest node;

[0083] Step S206: monitoring and maintenance, the specific steps of the operation are:

[0084] First, monitor the performance of each device node, and its performance set is:

[0085] ;

[0086] in, For Node A set of monitoring indicators, For load, For data access latency, is the storage capacity;

[0087] Then, the maintenance task of the node is performed regularly, the formula is:

[0088] ;

[0089] in, For Node Maintenance operations, It is a periodic maintenance task function.

[0090] Further, the step S3 specifically includes:

[0091] Step S301: Clean, convert and standardize the collected data to eliminate noise, missing values ​​and inconsistency in the data to ensure data quality;

[0092] Step S302: Model and analyze the preprocessed time series data using a time series model to mine and predict the time laws in urban operation, including but not limited to the time series change trends of traffic flow, energy consumption, and population flow. This step includes the following specific steps:

[0093] First, determine the parameters of the time series model, including the number of autoregressive terms , sliding average number of items and the difference order ;

[0094] Then, the parameters are estimated according to the time series model formula to obtain the model parameters and , the timing model formula is:

[0095] ;

[0096] in, is the coefficient of the autoregressive term, which determines the impact of historical data on the current value. The number of autoregressive terms p is the highest order of the autoregressive part in the model. is the backshift operator, defined as ,in is the first-order autoregressive coefficient of the original sequence, is the high-order autoregressive coefficient of the original sequence, Backward Operator of The power represents the original sequence. order difference is the highest order of the difference part in the model, Backward Operator of Power, indicating a point in time The error term Before The relationship between the error terms at each time point is Backward Operator of Power, used to describe Before The relationship between the values ​​at each time point, is the coefficient of the sliding average term, which determines the impact of the moving average process on the current value. is the highest order of the sliding average part in the model, is the error term, representing the random fluctuation or noise of the model, is the observed value of the time series data at time point t, The calculation expression is:

[0097] ;

[0098] in, represents a constant term, For time series data at time points The observed value of is the coefficient of the autoregressive term, Indicates at time The error term or random disturbance of represents the coefficient of the sliding average term;

[0099] Furthermore, the estimated model parameters are used to predict the time series data to obtain the short-term prediction trend and long-term prediction trend of the equipment collection data;

[0100] Finally, based on the forecast results, analyze the time patterns of urban operation, including but not limited to periodicity, seasonality, and trend;

[0101] Step S303: Use the intrinsic association mining algorithm to discover the intrinsic relationship between different data to provide a basis for model construction.

[0102] Furthermore, in step S303, the step of using the intrinsic association mining algorithm to discover the intrinsic relationship between different data is:

[0103] (1) Define the minimum correlation and minimum trust ,According to the specific needs and data characteristics of the city twin system, the minimum correlation and minimum trust thresholds are set to filter frequent pattern sets and generate effective correlation relationships;

[0104] Among them, the correlation Representing data with intrinsic relationships and The number of times it appears in the data set accounts for the total amount of the data set The proportion of , The corresponding correlation is expressed as:

[0105] ;

[0106] Trust For data After it appears, other data The probability of occurrence is defined as:

[0107] ;

[0108] in, Representation data and The number of times it appears at the same time, Representation data and Probability of simultaneous occurrence; Represents the occurrence data Y The probability of Representative data Y degree of relevance;

[0109] (2) Generate candidate item sets: Starting from a candidate item set containing only one data item, gradually build a candidate item set containing multiple data items, and use the prior principle to perform pruning to reduce the amount of computation;

[0110] (3) Calculate the association and filter frequent itemsets: For each candidate itemset, calculate its frequency of occurrence in all transactions, i.e., the association; and compare it with the minimum association threshold to filter out all frequent itemsets, where Representative data The correlation degree of , the screening method is:

[0111] ;

[0112] (4) Calculate the trust and generate the association relationship: For each pair of frequent itemsets and ,in yes A superset of If the trust is greater than or equal to the minimum trust threshold, the rule is considered a valid association relationship. The conditions for determining a valid association relationship are:

[0113] ;

[0114] (5) Discover the intrinsic connections between different data by mining the generated associations;

[0115] (6) Optimize the city twin system based on the mined associations;

[0116] (7) According to the actual needs of urban operation, adjust the parameters of the intrinsic association mining algorithm to improve the accuracy of mining;

[0117] (8) Repeat steps (1) to (7) until the internal connection mining results of all cities are obtained.

[0118] Further, the step S4 specifically includes:

[0119] Step S401: The method of using a drone to collect data includes the following specific steps:

[0120] (1) According to the traditional acquisition method, the flight altitude is set to , set the heading overlap to , the lateral overlap is , collect fixed routes;

[0121] (2) Adjust the drone’s flight altitude and camera angle to manually shoot the area of ​​interest;

[0122] (3) Preprocess the images obtained by UAV oblique photography, further perform aerial triangulation, and obtain an aerial triangulation quality report;

[0123] (4) Analyze the number of key points and connection points in the quality report and make a preliminary judgment on its accuracy;

[0124] (5) Then determine whether the reprojection error meets the accuracy requirements. If so, perform oblique photography to simultaneously generate a sparse point cloud and a preliminary constructed three-dimensional model;

[0125] Among them, the sparse point cloud is used to align with the ground laser radar point cloud data, obtain the city model accuracy from the initially constructed 3D model, and determine the measurement area of ​​the ground laser radar;

[0126] Step S402: using a ground-based laser radar to collect data and perform point cloud registration to obtain registered laser radar point cloud data;

[0127] Step S403: importing the oblique photography image of the drone and the registered laser radar point cloud data into the 3D modeling software at the same time, and using the software to generate a 3D model of the real city scene;

[0128] Step S404: Using machine learning methods based on the oblique photography images from drones to identify different communities or commercial areas in the city, and constructing refined models of various areas in the city.

[0129] Furthermore, in step S404, the steps of using machine learning methods to identify different communities or commercial areas in the city based on the oblique photography images of the drone and constructing refined models of various areas of the city are as follows:

[0130] First, a convolutional neural network is used to extract features from the preprocessed image data. The specific method is as follows:

[0131] ;

[0132] in, For the The convolution kernels are located at The output, and is the loop variable used to iterate in the convolution operation, For the input image at position The pixel value of For the The convolution kernels are located at The weight of is the size of the convolution kernel, For the The bias term of the convolution kernel; To calculate the distance from the center to the edge of the filter;

[0133] Then, the extracted feature vectors are classified to obtain the category labels of the objects. In the classification task, the output of the fully connected layer is converted into a probability distribution using the following function, which is:

[0134] ;

[0135] in, is an exponential function, is a constant with a natural base, Output vector for the fully connected layer No. elements, is the total number of categories, For the prediction Probability of class;

[0136] Finally, based on the classification results, a refined model of each area of ​​the city is constructed.

[0137] Further, the step S5 specifically includes:

[0138] Step S501: Integrate virtual reality VR and augmented reality AR technologies to create a visualization environment of a three-dimensional city model;

[0139] Step S502: inputting the city refined model data constructed in step S401 into the VR and AR system;

[0140] Step S503: Realize the real-time simulation of the city operation status. The traffic flow realization formula is as follows:

[0141] ;

[0142] in, Indicates time traffic flow, is the total number of road segments in the transportation network, For time Road Section The vehicle volume flow, For road section The time derivative of vehicle volume flow; the realization formula of pedestrian flow density is as follows:

[0143] ;

[0144] in, For the location and time The population density, Indicates at location and time Number of people, For location the area of ​​the region;

[0145] Step S504: Use VR and AR technology to display the real-time simulation results of the city operation status, including but not limited to traffic flow, crowd distribution, and facility operation status. The formula is as follows:

[0146] ;

[0147] in, For the visual presentation of simulation results, is the rendering function, for The city's operating status over time, It is the user’s perspective or observation point;

[0148] Step S505: According to user interaction and feedback, adjust simulation parameters to optimize simulation effect and user experience.

[0149] Further, the step S6 specifically includes:

[0150] Step S601: Initialize the online learning model and define the learning rate as , the model parameters are The initial state of the model is ;

[0151] ;

[0152] in, To initialize the online learning model function, for different online learning models;

[0153] Step S602: During the system operation, real-time input data Perform preprocessing, where t represents the time step; Represents the function of preprocessing the input data;

[0154] ;

[0155] Step S603: Apply online learning algorithm to update model parameters , in order to improve the system's adaptability to new data patterns, the update formula is as follows:

[0156] ;

[0157] in, represents the updated model parameters, represents the current model parameters, represents the learning rate, Represents the loss function Corresponding to the parameter In the data The gradient on

[0158] Step S604: Calculate the loss function , to evaluate the performance of the model under the current data mode, the loss function uses the mean square error metric:

[0159] ;

[0160] in, Indicated in the parameter and data The loss of represents the number of data points, represents the true value, For the prediction Probability of class;

[0161] Step S605: Adjust learning rate based on model performance and system feedback , to improve learning efficiency and system adaptability:

[0162] ;

[0163] in, represents the original learning rate, represents the adjusted learning rate; Represents the learning rate adjustment function; Indicates at time system performance indicators.

[0164] The beneficial effects of the present invention include:

[0165] (1) The data collection work of the present invention is based on the idea of ​​real-time monitoring. First, a data collection network is constructed, and the layout and collection frequency of various sensors and IoT devices are optimized by using the particle swarm optimization algorithm and the adaptive sampling algorithm. This not only ensures the real-time collection of urban operation data, but also optimizes the equipment layout scheme by using the particle swarm optimization algorithm, and intelligently adjusts the collection strategy by using the adaptive sampling algorithm, thereby reducing the redundancy and error of data collection and providing a reliable data basis for subsequent urban operation analysis.

[0166] (2) The present invention proposes an efficient data storage and management method: using the cloud computing platform, a distributed database based on the consistent hashing algorithm is built, which realizes the uniform distribution and efficient access of data in the distributed database, which not only improves the efficiency of data storage, but also ensures the efficiency and stability of data management through the load balancing algorithm, and provides effective data storage and management for subsequent data mining;

[0167] (3) The present invention uses big data analysis technology to pre-process the collected data, uses the time series model to mine the time law of urban operation, and discovers the internal connection between different data through the internal association mining algorithm. This method not only reduces the complexity of data pre-processing, but also provides strong data support for model construction, thereby improving the accuracy and practicality of the model;

[0168] (4) The present invention proposes a method for fusion modeling of urban 3D real scenes, which mainly uses drone oblique photography and ground-based laser radar to build a city model, and uses machine learning methods to identify different communities or commercial areas in the city, and constructs a refined model of various areas in the city. This method not only improves the precision of the model, but also enhances the model's ability to identify different urban areas, making the constructed 3D city model more accurate;

[0169] (5) The present invention realizes real-time simulation of urban operation status through technologies such as virtual reality and augmented reality, which not only provides intuitive decision-making support for urban management, but also significantly enhances the comprehensive response capability of cities to cope with emergencies by simulating different urban operation scenarios, thereby laying a solid technical foundation for promoting sustainable urban development and is expected to play a key role in future urban planning and emergency management systems;

[0170] (6) The present invention is based on an online learning algorithm to continuously optimize the system, so that the system can continuously learn and adapt to new data patterns during operation. This method not only improves the construction and operation optimization capabilities of the digital twin city, but also ensures that the system can maintain efficient operation and provide accurate results in the face of a constantly changing urban environment, ultimately improving the construction and operation optimization capabilities of the digital twin city.

[0171] Other advantages, objectives and features of the present invention will be described in detail in the subsequent description. Through in-depth study of the following, those skilled in the art will be able to clearly recognize these advantages and features and obtain valuable teachings from the practice of the present invention. The objectives and other advantages of the present invention can be realized and reflected in the following description and the aforementioned claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0172] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0173] Figure 1 This is a flow chart of a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology of the present invention;

[0174] Figure 2 A flow chart of optimizing the layout of various sensors and IoT devices based on the particle swarm optimization algorithm of the present invention;

[0175] Figure 3 This is an example diagram of a method for combining automatic route photography and manual adjustment photography of the present invention;

[0176] Figure 4 This is an example diagram of the distribution of the location points of the ground laser radar stations of the present invention;

[0177] Figure 5 The technical roadmap of three-dimensional city modeling of the present invention;

[0178] Figure 6 , Figure 7 This is an example diagram of the practical application of the urban digital twin system of the present invention. DETAILED DESCRIPTION

[0179] In order to further explain the technical means and effects taken by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. It should be clear that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0180] like Figure 1 As shown, a digital twin city construction and operation optimization method based on cloud computing and big data technology of the present invention comprises the following steps:

[0181] Step S1: Build a data collection network, optimize the layout of various sensors and IoT devices based on the particle swarm optimization algorithm and adaptive sampling algorithm, adjust the sampling frequency, and collect urban operation data in real time;

[0182] Step S2: Using the cloud computing platform to build a distributed database, using a load balancing algorithm to evenly distribute data in the distributed database, and achieving efficient data storage and management;

[0183] Step S3: pre-process the collected data based on big data analysis technology, use the time series model to mine the time law of urban operation and use the intrinsic association mining algorithm to find the association between different data, so as to provide a basis for model construction;

[0184] Step S4: construct a city model by using a 3D real-scene fusion modeling method based on drone oblique photography and supplemented by ground-based laser radar, apply machine learning methods to identify different communities or commercial areas in the city, and construct refined models of various areas in the city;

[0185] Step S5: Realize real-time simulation of city operation status through virtual reality and augmented reality technology;

[0186] Step S6: Use online learning algorithms to continuously optimize the system so that the system can continuously learn and adapt to new data patterns during operation, thereby improving the construction and operation optimization capabilities of the digital twin city.

[0187] It should be noted that any process or method description in the flowchart of the present invention or described in other ways herein may be understood to represent a module, fragment or portion of a code including one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations, in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present invention belong.

[0188] like Figures 2 to 7 As shown, the specific steps of the above method will be further described below through a specific embodiment.

[0189] The step S1 comprises the following sub-steps:

[0190] Step S101: Establish the deployment goals of sensors and IoT devices and define the objective function Used to evaluate the performance of layout solutions, where The parameter vector representing the equipment layout, the objective function calculation formula is:

[0191] ;

[0192] in, is the coverage rate, is the device density, is the acquisition frequency, , , are the corresponding weight coefficients; in this embodiment, they are 0.4, 0.3, and 0.3 respectively;

[0193] Step S102: Optimizing the layout of sensors and IoT devices using a particle swarm optimization algorithm;

[0194] In step S102, the specific steps of optimizing it using the particle swarm optimization algorithm are as follows:

[0195] (1) Initialize a group of particles, each particle represents a device layout scheme, where i=1, 2, ..., In this embodiment, The value is 100;

[0196] (2) Evaluate the fitness of each particle , i.e., the performance of the layout scheme;

[0197] (3) Update particle speed and equipment layout plan , the updated calculation formula is:

[0198] ;

[0199] ;

[0200] in, for The speed of the time particle, for Time equipment layout plan, for Time equipment layout plan, is the inertia weight, which is 0.7 in this embodiment; and is the learning factor, with values ​​of 0.5 and 0.3 respectively; and is a random number between [0, 1]; is the optimal position of an individual, For the optimal layout solution, is the disturbance coefficient, and the value of this embodiment is 0.66; For the The dynamic disturbance vector of a particle, ={0.1, -0.1, 0.23, 0.65}; is the position update coefficient, which is set to 0.4 in this embodiment; for The speed of time particles;

[0201] (4) Repeat the above steps until the termination condition is met. = , is the termination condition time. In this embodiment, The value is 100, and the optimal layout solution is obtained. ;

[0202] Step S103: Obtaining the optimal layout solution Finally, an adaptive sampling algorithm is used to adjust the acquisition frequency of the device.

[0203] In step S103, the specific steps of the adaptive sampling algorithm to adjust the acquisition frequency of the device are:

[0204] First, according to real-time data changes and preset thresholds , adaptive adjustment Time Single Device Collection Frequency , the adaptive adjustment formula is:

[0205] ;

[0206] in, for Time single device collection frequency, is the adjustment coefficient, and the value of this embodiment is 0.1; To adjust the index, is a decay term that increases with time, where is the attenuation coefficient, which is 0.86 in this embodiment; is the time step, and the decay term will change with time increases with the increase of The impact of

[0207] The adjustment formula for the substituted values ​​is:

[0208] ;

[0209] Among them, the acquisition frequency adjustment factor of each device The calculation formula is:

[0210] ;

[0211] in, For the The optimal acquisition frequency for each device is 10 times per minute in this embodiment. For the The current collection frequency of each device, which is 5 times per minute in this embodiment;

[0212] Then, design Objective function, find its minimum value, used for evaluation performance;

[0213] ;

[0214] in, is the coverage rate, which is 0.88 in this embodiment, indicating the proportion of urban areas that can be covered by the acquisition frequency; The device load is set to 60%, which indicates the workload of the device under the set acquisition frequency. The value range of this workload is determined by the physical device and does not exceed the maximum workload. , , are corresponding weight coefficients used to balance the importance of different objectives, and in this embodiment, they are 0.88, 0.65, and 0.89 respectively; is the data value, which indicates the importance of the collected data to urban management and decision-making, and its range is [-1, 1]; It is expressed as:

[0215] ;

[0216] in, , , They represent the impact of data on decision making, the real-time nature of data, and the integrity of data, respectively, and their value range is [0, 1]; , , Respectively represent their weight coefficients, with values ​​of 0.8, 0.6, and 0.6;

[0217] Finally, real-time collection of urban operation data ensures the real-time and accuracy of the data and provides basic data support for subsequent analysis and simulation. The data collection process can be expressed as:

[0218] ;

[0219] ;

[0220] in, For the Devices at time The collected data, For single Devices at time The collected data, For time Single device acquisition frequency, For time A collection of data collected from different devices.

[0221] The step S2 specifically includes the following steps:

[0222] Step S201: Design a distributed database architecture. The specific steps are as follows:

[0223] First, determine the node topology and assign a unique identifier, where the topology and identifier are calculated as follows:

[0224] ;

[0225] ;

[0226] in, is the node topology, For the nodes, is the total number of nodes, For Node A unique identifier for Compute a function for an identifier;

[0227] In this embodiment, there are 100 nodes, node 1, node 2, and node 3, etc., with a total number of nodes n=100, and unique identifiers are assigned; the ID1 of node 1 is 001, the ID2 of node 2 is 002, and the ID3 of node 3 is 003. The topology structure for:

[0228] ;

[0229] Then, the node hash value is calculated, where the calculation formula of the node hash value is:

[0230] ;

[0231] in, For Node The hash value of is a hash function, which is MD5 in this embodiment; is the number of bits of the hash ring, which is 10 in this embodiment; calculate the hash value of node 1:

[0232] ;

[0233] Step S202: Implement the consistent hashing algorithm. The specific steps are as follows:

[0234] First, initialize the hash ring, where the initialization formula is:

[0235] ;

[0236] in, is a hash ring;

[0237] Then, map the node to the hash ring and determine the node position, where the mapping calculation formula is:

[0238] ;

[0239] ;

[0240] in, For Node Position on the hash ring;

[0241] Step S203: Data distribution and storage: Map data items to hash rings and determine data storage locations. The formula is as follows:

[0242] ;

[0243] ;

[0244] in, For data items The position on the hash ring, For data items The hash value of To store data items Nodes; To find the node function;

[0245] Step S204: load balancing and data migration. The specific steps of the operation are:

[0246] First, the data migration when adding a new node is calculated as follows:

[0247] ;

[0248] in, Indicates the data migration of newly added nodes. is the hash value difference between the newly added node and the node where the data item is currently located, For data items The amount of migration;

[0249] Then, delete the data migration of the node. The calculation formula for deleting data migration is:

[0250] ;

[0251] in, is the hash value difference between the deleted node and the node where the data item currently resides;

[0252] Step S205: data access and query optimization. The specific steps of the operation are:

[0253] First, build an index of device collection data for query optimization. The method is as follows:

[0254] ;

[0255] ;

[0256] in, For data items The index of Build a function for the index, Used to calculate query time, Used to obtain the index size;

[0257] Then, the cache strategy for collected data is:

[0258] ;

[0259] ;

[0260] in, For data items The cache status, is the time step, is the cache strategy function, for The number of cache accesses, for Total visits, Indicates the ratio of the number of successful data acquisitions from the cache to the total number of data acquisition requests within a set time (set to 5 seconds in this embodiment), which is an indicator used to measure cache performance;

[0261] Finally, the query routing process of the device collection data, the query expression is:

[0262] ;

[0263] in, To query the routing function, A function to find the nearest node;

[0264] Step S206: monitoring and maintenance. The specific steps of the operation are:

[0265] First, monitor the performance of each device node, and its performance set is:

[0266] ;

[0267] in, For Node A set of monitoring indicators, For load, For data access latency, is the storage capacity;

[0268] Then, perform maintenance tasks regularly, the formula is as follows:

[0269] ;

[0270] in, For Node Maintenance operations, It is a periodic maintenance task function.

[0271] The step S3 specifically includes the following steps:

[0272] Step S301: Clean, convert and standardize the collected traffic flow, air quality and population density data to eliminate noise, missing values ​​and inconsistency in the data and ensure data quality. The specific steps in the data preprocessing process are:

[0273] (1) Missing value filling: The mean filling method is used. Since there are many missing values ​​in the traffic flow data, the mean of the first 100 data is used for filling;

[0274] (2) Standardization: Use the formula Transform the data to a standard normal distribution, where is the original data, is the mean, is the standard deviation;

[0275] Step S302: Model and analyze the preprocessed time series data using a time series model to mine and predict the time laws in urban operation, including but not limited to the time series change trends of traffic flow, energy consumption, and population flow. This step includes the following specific steps:

[0276] First, determine the parameters of the time series model, including the number of autoregressive terms , its value is 2; the number of sliding average items , its value is 1; the difference order , its value is 1;

[0277] Then, the parameters are estimated according to the time series model formula to obtain the model parameters and , the timing model formula is:

[0278] ;

[0279] in, is the coefficient of the autoregressive term, which determines the impact of historical data on the current value. is the highest order of the autoregressive part of the model, is the backshift operator, defined as ,in is the first-order autoregressive coefficient of the original sequence, is the high-order autoregressive coefficient of the original sequence, Backward Operator of The power represents the original sequence. order difference is the highest order of the difference part in the model, Backward Operator of Power, indicating a point in time The error term Before The relationship between the error terms at each time point is Backward Operator of Power, used to describe Before The relationship between the values ​​at each time point, is the coefficient of the sliding average term, which determines the impact of the moving average process on the current value. is the highest order of the sliding average part in the model, is the error term, representing the random fluctuation or noise of the model, is the observed value of the time series data at time point t, The calculation expression is:

[0280] ;

[0281] in, represents a constant term, For time series data at time points The observed value of is the coefficient of the autoregressive term, Indicates at time The error term or random disturbance of represents the coefficient of the sliding average term;

[0282] Furthermore, the estimated model parameters are used to predict the time series data to obtain the short-term prediction trend and long-term prediction trend of the equipment collection data;

[0283] Finally, based on the forecast results, analyze the time patterns of urban operation, including but not limited to periodicity, seasonality, and trend;

[0284] Step S303: using the intrinsic association mining algorithm to discover the intrinsic relationship between different data, providing a basis for model construction. The steps of using the intrinsic association mining algorithm to discover the intrinsic relationship between different data are as follows:

[0285] (1) Define the minimum correlation , in this embodiment, its value is 0.1; minimum trust , in this embodiment, its value is 0.7. According to the specific requirements and data characteristics of the urban digital twin system, the minimum correlation and minimum trust thresholds are set to filter frequent pattern sets and generate effective association relationships;

[0286] Among them, the correlation Representing data with intrinsic relationships and The number of times it appears in the data set accounts for the total amount of the data set The proportion of , The corresponding correlation is expressed as:

[0287] ;

[0288] Trust For data After it appears, other data The probability of occurrence is defined as:

[0289] ;

[0290] in, Representation data and The number of times it appears at the same time, Representation data and Probability of simultaneous occurrence; Represents the occurrence data Y The probability of Representative data Y degree of relevance;

[0291] (2) Generate candidate item sets: Starting from a candidate item set containing only one data item (such as “high traffic flow”), gradually build a candidate item set containing multiple data items and use a priori principles to perform pruning to reduce the amount of computation;

[0292] (3) Calculate the association and filter frequent itemsets: For each candidate itemset, calculate its frequency of occurrence in all transactions (i.e., the association), and compare it with the minimum association threshold to filter out all frequent itemsets. In this embodiment, Representative data The correlation degree of , the screening method is:

[0293] ;

[0294] (4) Calculate the trust and generate the association relationship: For each pair of frequent itemsets and ,in yes A superset of If the trust is greater than or equal to the minimum trust threshold, the rule is considered a valid association rule;

[0295] Among them, in this embodiment, the effective association rule judgment condition is:

[0296] ;

[0297] (5) Discover the intrinsic connections between different data by mining the generated associations. For example, the rule “high traffic flow → high air pollution” may indicate that traffic congestion is an important factor leading to deterioration of air quality.

[0298] (6) Optimize the urban twin system based on the mined correlations. For example, based on the correlation between traffic flow and air quality, the control strategy of traffic lights can be adjusted to reduce congestion and air pollution.

[0299] (7) According to the actual needs of urban operation, adjust the parameters of the intrinsic association mining algorithm to improve the accuracy of mining;

[0300] (8) Repeat steps (1) to (7) until the internal connection mining results of all city operations are obtained;

[0301] The step S4 specifically includes the following steps:

[0302] Step S401: The specific steps of the method for collecting data using a drone are as follows:

[0303] (1) Set the flight altitude according to the traditional acquisition method =110 m, set the heading overlap =75%, lateral overlap =65%, collect fixed routes;

[0304] (2) Adjust the drone’s flight altitude and camera angle to manually shoot the area of ​​interest;

[0305] (3) The images obtained by UAV oblique photography need to be preprocessed first, and then aerial triangulation is performed to obtain an aerial triangulation quality report;

[0306] (4) Analyze the number of key points and connection points in the quality report and make a preliminary judgment on its accuracy;

[0307] (5) Then determine whether the reprojection error meets the accuracy requirements. If so, perform oblique photography to simultaneously generate a sparse point cloud and a preliminary constructed three-dimensional model;

[0308] Among them, the sparse point cloud is used to align with the ground laser radar point cloud data, obtain the city model accuracy from the initially constructed 3D model, and determine the measurement area of ​​the ground laser radar;

[0309] Step S402: Collect data using ground laser radar data:

[0310] (1) First, determine the survey area and arrange the survey stations according to the scope of the survey area. The scanning range of the ground laser radar Leica BLK360 used in this embodiment is 0.6~60 m. The farther the distance of the scanning point is, the lower the point cloud accuracy and ranging accuracy are. The actual scanning distance of a survey station is controlled between 20~30 m. In order to facilitate the intelligent stitching of the point cloud, there needs to be a degree of overlap between adjacent survey stations to ensure that adjacent survey stations can be connected;

[0311] (2) Use the point cloud data processing software Leica Cyclone Register360 to export the point cloud data and crop it to retain only the data within 20 m. This can effectively remove noise points and invalid points in the data and improve the subsequent data processing speed;

[0312] (3) Since the point cloud obtained by the ground-based laser radar has a high density, it needs to be downsampled to reduce the amount of point cloud data and improve the efficiency of subsequent point cloud processing. The voxel filter divides the input point cloud data into tiny three-dimensional voxel grids, and uses the center of gravity of the voxel grid to approximately represent other points. The voxel filter can reduce the amount of point cloud data without destroying the geometric structure of the point cloud data itself.

[0313] (4) Use statistical filters to remove noise points from the point cloud dataset. The statistical filter is used to calculate the average distance from each point to its K adjacent points. It is known that the result conforms to the Gaussian distribution, and its shape is determined by the mean and standard deviation of the average distance. Points with average distances outside the standard range can be defined as outliers and removed from the data.

[0314] (5) A combination of manual coarse registration and automatic fine registration is used for point cloud registration. Fine registration uses a point cloud matching algorithm, and its registration effect is closely related to the initial position of the point set. It cannot be directly used for point cloud fine registration. First, a manual coarse registration method is used to quickly narrow the matching search area and determine the point cloud set to be matched. Then, a point cloud matching algorithm is applied for automatic fine registration. The point cloud matching algorithm uses corresponding points to make the objective function Minimize, the calculation formula is as follows:

[0315] ;

[0316] in, is the number of iterations, which is 100 in this embodiment; i is a point in the reference point cloud; is the corresponding point in the target point cloud; and is the calculated rotation and translation matrix; is the weight coefficient, which is 0.33 in this embodiment;

[0317] Step S403, importing the oblique photography image of the drone and the registered laser radar point cloud data into Context Capture 3D modeling software at the same time, and using the software to generate a real-scene 3D model of the city;

[0318] Step S404: using machine learning methods based on oblique photography images from drones to identify different communities or commercial areas in the city, and constructing refined models of various areas in the city;

[0319] First, a convolutional neural network is used to extract features from the preprocessed image data. The specific method is as follows:

[0320] ;

[0321] in, For the The convolution kernels are located at The output, and is the loop variable used to iterate in the convolution operation, For the input image at position The pixel value of For the The convolution kernels are located at The weight of is the size of the convolution kernel (usually an odd number so that it has a center), For the The bias term of the convolution kernel; To calculate the distance from the center of the filter to the edge; in this embodiment, the size of the convolution kernel =3, the number of convolution kernels =10, the pixel value of the input image It is 128×128 image data;

[0322] Then, the extracted feature vectors are classified to obtain the category labels of the objects. In the classification task, the softmax function is used to convert the output of the fully connected layer into a probability distribution, and its formula is:

[0323] ;

[0324] in, is an exponential function, is a constant with a natural base, Output vector for the fully connected layer No. elements, is the total number of categories, For the prediction Probability of class;

[0325] Finally, based on the classification results, a refined model of each area of ​​the city is constructed;

[0326] The step S5 specifically includes the following steps:

[0327] Step S501: Integrate virtual reality VR and augmented reality AR technologies to create a visualization environment of a three-dimensional city model;

[0328] Step S502: input the city refined model data constructed in step S401 into the VR and AR systems to ensure the accuracy and completeness of the model data;

[0329] Step S503: Realize the real-time simulation of the city operation status. The specific implementation formula is as follows: ;

[0330] in, Indicates time traffic flow, is the total number of road segments in the transportation network, For time Road Section The vehicle volume flow, For road section The time derivative of vehicle volume flow; the realization formula of pedestrian flow density is as follows:

[0331] ;

[0332] in, For the location and time The population density, Indicates at location and time Number of people, For location the area of ​​the region;

[0333] Step S504: Use VR and AR technology to display the real-time simulation results of the city operation status, including but not limited to traffic flow, crowd distribution, facility operation status, etc. The formula is as follows:

[0334] ;

[0335] in, For the visual presentation of simulation results, is the rendering function, for The city's operating status over time, It is the user’s perspective or observation point;

[0336] Step S505: adjusting simulation parameters according to user interaction and feedback to optimize simulation effect and user experience;

[0337] Step S6 specifically includes the following steps:

[0338] Step S601: Initialize the online learning model and define the learning rate as , in this embodiment, it is taken as 0.01, and the model parameters are The initial state of the model is ;

[0339] ;

[0340] in, To initialize the online learning model function, For different online learning models, the FTRL (Follow The Regularized Leader) model is selected in this embodiment;

[0341] Step S602: During the system operation, real-time input data Perform preprocessing, where t represents the time step; Represents the function of preprocessing the input data;

[0342] ;

[0343] Step S603: Apply online learning algorithm to update model parameters , in order to improve the system's adaptability to new data patterns, the update formula is as follows:

[0344] ;

[0345] in, represents the updated model parameters, represents the current model parameters, represents the learning rate, Represents the loss function Corresponding to the parameter In the data The gradient on

[0346] Step S604: Calculate the loss function , to evaluate the performance of the model under the current data mode, the loss function uses the mean square error metric:

[0347] ;

[0348] in, Indicated in the parameter and data The loss of represents the number of data points, represents the true value, For the prediction Probability of class;

[0349] Step S605: Adjust learning rate based on model performance and system feedback , to improve learning efficiency and system adaptability:

[0350] ;

[0351] in, represents the original learning rate, represents the adjusted learning rate; Represents the learning rate adjustment function; Indicates at time system performance indicators.

[0352] In this embodiment, Figure 2 The flow chart of the present invention is presented to optimize the layout of various sensors and IoT devices based on the particle swarm optimization algorithm, which is used to build a data acquisition network; Figure 3 An example diagram showing the drone data collection method is a combination of automatic route shooting and manual adjustment shooting; Figure 4 This is an example diagram of the distribution of ground lidar stations; Figure 5 A technical roadmap for 3D city modeling by integrating oblique photography technology with ground-based LiDAR technology; Figure 6 and Figure 7 The practical application example diagrams of the city digital twin system of the present invention are respectively shown. The present invention improves the working efficiency of the digital twin city system based on cloud computing and big data technology through efficient processing algorithms and data acquisition mechanisms, which helps to promote the informatization, intelligence and green development of cities and improve the level of modernization of urban governance.

[0353] Considering the limitations of the traditional digital twin city system establishment method, the present invention makes full use of advanced optimization algorithms and big data analysis technologies, and uses all-round image data preprocessing and efficient target recognition algorithms to clean and classify the data collected by different types of Internet of Things devices, build a cloud computing platform and build a distributed database to efficiently store and manage the data. In the data analysis link, the present invention makes full use of big data analysis technology to preprocess the collected data, uses the time series model to mine the time law of urban operation, and discovers the intrinsic connection between different data through the intrinsic association mining algorithm. In terms of establishing a three-dimensional model of the city, the present invention adopts a method based on drone oblique photography and ground laser radar as a supplement, combines machine learning technology to identify different communities or commercial areas in the city, and constructs a refined model of various fields in the city. Furthermore, through technologies such as virtual reality and augmented reality, the present invention realizes real-time simulation of the city's operating status. Finally, based on the online learning algorithm, the present invention continuously optimizes the system, so that the system can continuously learn and adapt to new data patterns during operation, and improves the construction and operation optimization capabilities of the digital twin city.

[0354] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0355] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology, characterized in that: The following steps are involved: Step S1: construct a data collection network, optimize the layout of various sensors and IoT devices based on the particle swarm optimization algorithm and the adaptive sampling algorithm, adjust the sampling frequency, and collect urban operation data in real time; the step S1 specifically includes: Step S101: Establish the deployment goals of sensors and IoT devices and define the objective function Used to evaluate the performance of layout solutions, where The parameter vector representing the equipment layout, the objective function calculation formula is: ； in, is the coverage rate, is the device density, is the acquisition frequency, is the corresponding weight coefficient; Step S102: Optimizing the layout of sensors and IoT devices using a particle swarm optimization algorithm. The specific steps of optimizing the layout using the particle swarm optimization algorithm are as follows: (1) Initialize a group of particles, each particle represents a device layout scheme, where i=1, 2, ..., ; (2) Evaluate the fitness of each particle , i.e., the performance of the layout scheme; (3) Update particle speed and equipment layout plan , the updated calculation formula is: ； ； in, for The speed of the time particle, for Time equipment layout plan, for Time equipment layout plan, is the inertia weight, and is the learning factor, and is a random number between [0, 1], is the optimal position of an individual, For the optimal layout solution, is the disturbance coefficient, For the The dynamic disturbance vector of a particle, is the position update coefficient; for The speed of time particles; (4) Repeat the above steps until the termination condition is met. = ,in is the termination condition time, and the optimal layout solution is obtained ; Step S103: Using the optimal layout solution ,Adopt an adaptive sampling algorithm to adjust the acquisition frequency of the device; In step S103, the steps of using the adaptive sampling algorithm to adjust the acquisition frequency of the device are: First, according to real-time data changes and preset thresholds , adaptive adjustment Time Single Device Collection Frequency , the adaptive adjustment formula is: ； in, for Time single device collection frequency, is the adjustment factor, To adjust the index, is a decay term that increases with time, where is the attenuation coefficient, is the time step, and the decay term will change with time increases with the increase of The impact of Among them, the acquisition frequency adjustment factor of each device The calculation formula is: ； in, For the The optimal acquisition frequency for each device, For the The current acquisition frequency of each device; Then, design Objective function, find its minimum value, used for evaluation performance; ； in, is the coverage rate, which indicates the proportion of urban areas that can be covered by the acquisition frequency; The device load represents the workload of the device at a specific acquisition frequency. The value range of this workload is determined by the physical device and does not exceed the maximum workload. , , is the corresponding weight coefficient, which is used to balance the importance of different objectives; Data value, which indicates the importance of the collected data to urban management and decision-making; It is expressed as: ； in, , , They represent the impact of data on decision making, the real-time nature of data, and the integrity of data, respectively, and their value range is [0, 1]. , , Represent their weight coefficients respectively; Finally, real-time collection of urban operation data ensures the real-time and accuracy of the data and provides basic data support for subsequent analysis and simulation. The data collection process is expressed as: ； ； in, For the Devices at time The collected data, For single Devices at time The collected data, For time The acquisition frequency, For time The collection of data collected from different devices; Step S2: Using the cloud computing platform to build a distributed database, using a load balancing algorithm to evenly distribute data in the distributed database, and achieving efficient data storage and management; Step S3: pre-process the collected data based on big data analysis technology, use the time series model to mine the time law of urban operation and use the internal association mining algorithm to find the association between different data, so as to provide a basis for model construction; Step S4: construct a city model by using a 3D real-scene fusion modeling method based on drone oblique photography and supplemented by ground-based laser radar, apply machine learning methods to identify different communities or commercial areas in the city, and construct refined models of various areas in the city; Step S5: Realize real-time simulation of city operation status through virtual reality and augmented reality technology; Step S6: Use online learning algorithms to continuously optimize the system so that the system can continuously learn and adapt to new data patterns during operation, thereby improving the construction and operation optimization capabilities of the digital twin city.

2. According to claim 1, a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology is characterized by: The step S2 specifically includes: Step S201: Design a distributed database architecture. The specific steps are as follows: First, determine the node topology and assign a unique identifier. The calculation formula for the topology and identifier is: ； ； in, is the node topology, For the nodes, is the total number of nodes, For Node A unique identifier for Compute a function for an identifier; Then, the node hash value is calculated, where the calculation formula of the node hash value is: ； in, For Node The hash value of is a hash function, is the number of bits of the hash ring; Step S202: Implement the consistent hashing algorithm. The specific steps are as follows: First, initialize the hash ring, where the initialization formula is: ； in, is a hash ring; Then, map the node to the hash ring and determine the node position, where the mapping calculation formula is: ； ； in, For Node Position on the hash ring; Step S203: Data distribution and storage, mapping data items to hash rings and determining data storage locations, the formula is as follows: ； ； in, For data items The position on the hash ring, For data items The hash value of To store data items Nodes; To find the node function; Step S204: Load balancing and data migration. The specific steps of the operation are: First, the data migration when adding a new node is calculated as follows: ； in, Indicates the data migration of newly added nodes. is the hash value difference between the newly added node and the node where the data item is currently located, For data items The amount of migration; Then, delete the data migration of the node. The calculation formula for deleting data migration is: ； in, The hash value difference between the deleted node and the node where the data item is currently located; Step S205: Data access and query optimization. The specific steps of the operation are: First, build an index of device collection data for query optimization. The method is as follows: ； ； in, For data items The index of Build a function for the index, Used to calculate query time, Used to obtain the index size; Then, the cache strategy for collected data is: ； ； in, For data items The cache status, is the time step, is the cache strategy function, for The number of cache accesses, for Total visits, Indicates the ratio of the number of successful data acquisitions from the cache to the total number of data acquisition requests within a set time, which is used to measure cache performance. Finally, the query routing process of the device collection data, the query expression is: ； in, To query the routing function, A function to find the nearest node; Step S206: monitoring and maintenance, the specific steps of the operation are: First, monitor the performance of each device node, and its performance set is: ； in, For Node A set of monitoring indicators, For load, For data access latency, is the storage capacity; Then, the maintenance task of the node is performed regularly, the formula is: ； in, For Node Maintenance operations, It is a periodic maintenance task function.

3. According to claim 1, a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology is characterized in that: The step S3 specifically includes: Step S301: Clean, convert and standardize the collected data to eliminate noise, missing values ​​and inconsistency in the data to ensure data quality; Step S302: Model and analyze the preprocessed time series data using a time series model to mine and predict the time laws in urban operation, including the time series change trends of traffic flow, energy consumption, and population flow. This step includes the following specific steps: First, determine the parameters of the time series model, including the number of autoregressive terms , sliding average number of items and the difference order ; Then, the parameters are estimated according to the time series model formula to obtain the model parameters and , the timing model formula is: ； in, is the coefficient of the autoregressive term, which determines the impact of historical data on the current value. The number of autoregressive terms p is the highest order of the autoregressive part in the model. is the backshift operator, defined as ,in is the first-order autoregressive coefficient of the original sequence, is the high-order autoregressive coefficient of the original sequence, Backward Operator of The power represents the original sequence. order difference is the highest order of the difference part in the model, Backward Operator of Power, indicating a point in time The error term Before The relationship between the error terms at each time point is Backward Operator of Power, used to describe Before The relationship between the values ​​at each time point, is the coefficient of the sliding average term, which determines the impact of the moving average process on the current value. is the highest order of the sliding average part in the model, is the error term, representing the random fluctuation or noise of the model, is the observed value of the time series data at time point t, The calculation expression is: ； in, represents a constant term, For time series data at time points The observed value of is the coefficient of the autoregressive term, Indicates at time The error term or random disturbance of represents the coefficient of the sliding average term; Furthermore, the estimated model parameters are used to predict the time series data to obtain the short-term prediction trend and long-term prediction trend of the equipment collection data; Finally, based on the forecast results, analyze the time patterns of urban operation, including periodicity, seasonality, and trend; Step S303: Use the intrinsic association mining algorithm to discover the intrinsic relationship between different data to provide a basis for model construction.

4. According to claim 3, a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology is characterized in that: In step S303, the steps of using the intrinsic association mining algorithm to discover the intrinsic relationship between different data are: (1) Define the minimum correlation and minimum trust ,According to the specific needs and data characteristics of the city twin system, the minimum correlation and minimum trust thresholds are set to filter frequent pattern sets and generate effective correlation relationships; Among them, the correlation Representing data with intrinsic relationships and The number of times it appears in the data set accounts for the total amount of the data set The proportion of , The corresponding correlation is expressed as: ； Trust For data After it appears, other data The probability of occurrence is defined as: ； in, Representation data and The number of times it appears at the same time, Representation data and The probability of simultaneous occurrence, Represents the occurrence data The probability of Representative data degree of relevance; (2) Generate candidate item sets: Starting from a candidate item set containing only one data item, gradually build a candidate item set containing multiple data items, and use the prior principle to perform pruning to reduce the amount of computation; (3) Calculate the association and filter frequent itemsets: For each candidate itemset, calculate its frequency of occurrence in all transactions, i.e., the association; and compare it with the minimum association threshold to filter out all frequent itemsets, where Representative data The correlation degree of , the screening method is: ； (4) Calculate the trust and generate the association relationship: For each pair of frequent itemsets and ,in yes A superset of, calculation rules If the trust is greater than or equal to the minimum trust threshold, the rule is considered a valid association relationship. The conditions for determining a valid association relationship are: ； (5) Discover the intrinsic connections between different data by mining the generated associations; (6) Optimize the city twin system based on the mined associations; (7) According to the actual needs of urban operation, adjust the parameters of the intrinsic association mining algorithm to improve the accuracy of mining; (8) Repeat steps (1) to (7) until the internal connection mining results of all cities are obtained.

5. According to claim 1, a digital twin city construction and operation optimization method based on cloud computing and big data technology is characterized by: The step S4 specifically includes: Step S401: The method of using a drone to collect data includes the following specific steps: (1) Set the flight altitude according to the traditional acquisition method , set the heading overlap to , the lateral overlap is , collect fixed routes; (2) Adjust the drone’s flight altitude and camera angle to manually shoot the area of ​​interest; (3) Preprocess the images obtained by UAV oblique photography, further perform aerial triangulation, and obtain an aerial triangulation quality report; (4) Analyze the number of key points and connection points in the quality report and make a preliminary judgment on its accuracy; (5) Then determine whether the reprojection error meets the accuracy requirements. If so, perform oblique photography to simultaneously generate a sparse point cloud and a preliminary constructed three-dimensional model; Among them, the sparse point cloud is used to align with the ground laser radar point cloud data, obtain the city model accuracy from the initially constructed 3D model, and determine the measurement area of ​​the ground laser radar; Step S402: using a ground-based laser radar to collect data and perform point cloud registration to obtain registered laser radar point cloud data; Step S403: importing the oblique photography image of the drone and the registered laser radar point cloud data into the 3D modeling software at the same time, and using the software to generate a 3D model of the real city scene; Step S404: Using machine learning methods based on the oblique photography images from drones to identify different communities or commercial areas in the city, and constructing refined models of various areas in the city.

6. According to claim 5, a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology is characterized in that: In step S404, the steps of using machine learning methods to identify different communities or commercial areas in the city based on the oblique photography images of drones and constructing refined models of various areas of the city are as follows: First, a convolutional neural network is used to extract features from the preprocessed image data. The specific method is as follows: ； in, For the The convolution kernels are located at The output, and is the loop variable used to iterate in the convolution operation, For the input image at position The pixel value of For the The convolution kernels are located at The weight of is the size of the convolution kernel, For the The bias term of the convolution kernel; To calculate the distance from the center to the edge of the filter; Then, the extracted feature vectors are classified to obtain the category labels of the objects. In the classification task, the output of the fully connected layer is converted into a probability distribution using the following function, which is: ； Among them, among them, is an exponential function, is a constant with a natural base, Output vector for the fully connected layer No. elements, is the total number of categories, For the prediction Probability of class; Finally, based on the classification results, a refined model of each area of ​​the city is constructed.

7. According to claim 1, a method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology is characterized by: The step S5 specifically includes: Step S501: Integrate virtual reality VR and augmented reality AR technologies to create a visualization environment of a three-dimensional city model; Step S502: inputting the city refined model data constructed in step S401 into the VR and AR system; Step S503: Realize the real-time simulation of the city operation status. The traffic flow realization formula is as follows: ； in, Indicates time traffic flow, is the total number of road segments in the transportation network, For time Road Section The vehicle volume flow, For road section The time derivative of vehicle volume flow; the realization formula of pedestrian flow density is as follows: ； in, For the location and time The population density, Indicates at location and time Number of people, For location the area of ​​the region; Step S504: Use VR and AR technology to display the real-time simulation results of the city operation status, including traffic flow, crowd distribution, and facility operation status. The formula is as follows: ； in, For the visual presentation of simulation results, is the rendering function, for The city's operating status over time, It is the user’s perspective or observation point; Step S505: According to user interaction and feedback, adjust simulation parameters to optimize simulation effect and user experience.

8. The method for constructing and optimizing the operation of a digital twin city based on cloud computing and big data technology according to claim 1 is characterized by: The method further includes step S6, i.e., continuously optimizing the system using an online learning algorithm, so that the system can continuously learn and adapt to new data patterns during operation, thereby improving the construction and operation optimization capabilities of the digital twin city. Step S6 specifically includes: Step S601: Initialize the online learning model and define the learning rate as , the model parameters are The initial state of the model is ; ； in, To initialize the online learning model function, for different online learning models; Step S602: During the system operation, real-time input data Preprocessing is performed, in which represents the time step; Represents the function of preprocessing the input data; ； Step S603: Apply online learning algorithm to update model parameters , in order to improve the system's adaptability to new data patterns, the update formula is as follows: ； in, represents the updated model parameters, represents the current model parameters, represents the learning rate, Represents the loss function Corresponding to the parameter In the data The gradient on Step S604: Calculate the loss function , to evaluate the performance of the model under the current data mode, the loss function uses the mean square error metric: ； in, Indicated in the parameter and data The loss of represents the number of data points, represents the true value, For the prediction Probability of class; Step S605: Adjust learning rate based on model performance and system feedback , to improve learning efficiency and system adaptability: ； in, represents the original learning rate, represents the adjusted learning rate; Represents the learning rate adjustment function; Indicates at time system performance indicators.

Images