Patrol treatment platform for multi-dimensional data analysis

Abstract

The invention relates to the technical field of smart city patrol management, and provides a multi-dimensional data analysis patrol management platform, which comprises a data acquisition unit used for acquiring a multi-source security data set of a smart city; the data processing unit is used for preprocessing the multi-source security data set to obtain security multi-dimensional data; the data analysis unit is used for performing data analysis on the security multi-dimensional data to construct a region division model, performing region division on the smart city based on the region division model to obtain a region division result of the smart city, and the region division result comprises dividing the smart city into a plurality of regions, each region has a corresponding region management cost value, a region connection degree value, a region internal difference value and a region security risk value; and the patrol treatment unit is used for designing a patrol treatment scheme according to the region division result and managing the smart city based on the patrol treatment scheme. According to the method, region division of the smart city is realized, and multiple factors such as region management cost, region connection degree, region internal difference and safety risk are considered, so that more efficient and accurate management of the smart city is realized.

Description

Technical Field

[0001] The present invention relates to the field of smart city patrol and management technology, and in particular to a patrol and management platform for multi-dimensional data analysis. Background Art

[0002] With the continuous advancement of smart city construction, urban management faces increasingly complex challenges. In the process of urban modernization management, many problems and blind spots have been exposed. Through manual inspections and traditional equipment perception, it is difficult to achieve real-time and effective supervision of the "high, far, blocked, dark, moving, and dangerous" management blind spots. Furthermore, the traditional management model is relatively simple. When facing sudden emergency events, manual duty is difficult, response is slow, coordination is difficult, feedback efficiency is low, and it is easy to delay opportunities and cause losses to expand. Traditional inspection and governance methods often lack data support and are difficult to deal with large-scale, multi-dimensional urban safety issues.

[0003] When existing urban management platforms are put into operation, drones are often used to collect data for smart cities. However, the data collected by drones has time and space lags. When there is data missing, the data obtained through ordinary mean interpolation has large errors, and the analysis of the data to plan smart city patrol and management plans is not targeted enough. Summary of the invention

[0004] In view of this, the present invention proposes a patrol and management platform for multi-dimensional data analysis, which collects multi-source security data, pre-processes and analyzes the data in multiple dimensions, constructs a regional division model, realizes the regional division of smart cities, and designs targeted patrol and management plans based on the regional division results, taking into account multiple factors such as regional management costs, regional connection levels, internal differences in regions, and security risks, thereby achieving more efficient and accurate management of smart cities and solving the problem of poor targeting caused by missing data in patrol and management of smart cities.

[0005] The technical solution of the present invention is implemented as follows: The present invention provides a patrol management platform for multi-dimensional data analysis, including:

[0006] A data collection unit, used to collect multi-source security data sets of smart cities;

[0007] A data processing unit, used to pre-process the multi-source security data set to obtain secure multi-dimensional data;

[0008] A data analysis unit is used to perform data analysis on the security multi-dimensional data to construct a regional division model, divide the smart city into regions based on the regional division model, and obtain a regional division result of the smart city, wherein the regional division result includes dividing the smart city into multiple regions, and each region has a corresponding regional management cost value, a regional connection degree value, a regional internal difference value, and a regional security risk value;

[0009] The inspection and management unit is used to design inspection and management plans based on the regional division results and manage the smart city based on the inspection and management plans.

[0010] On the basis of the above technical solution, preferably, the pretreatment includes:

[0011] The multi-source security data set is standardized to obtain a multi-source security standard data set, the multi-source security standard data set is identified through a spatiotemporal data mining algorithm to mark the missing values ​​in the multi-source security standard data set, the missing values ​​are interpolated based on the spatiotemporal data association rules to obtain a multi-source security complete data set, and multi-modal data fusion is performed on the multi-source security complete data set to obtain secure multi-dimensional data.

[0012] On the basis of the above technical solution, preferably, the standardization includes:

[0013] Perform Z-score normalization on the multi-source security data set to make the multi-source security data on the same scale:

[0014]

[0015] in, is the original data of the multi-source security standard dataset, is the original data of the multi-source security dataset, is the mean of the multi-source security dataset, is the standard deviation of the multi-source security dataset.

[0016] Based on the above technical solution, preferably, identifying the multi-source safety standard data set through the spatiotemporal data mining algorithm includes:

[0017] Arrange the multi-source security standard data set in time order and space order to obtain a multi-source security space-time volume, and divide the multi-source security space-time volume into a plurality of cells, each cell including an equal number of data points;

[0018] Perform time series trend analysis on each cell to obtain the time missing score of each cell;

[0019] Perform spatial abnormal area analysis on each cell to obtain the area missing score of each cell;

[0020] The cells corresponding to the absolute values ​​of the regional missing scores that are greater than the preset values ​​are divided into abnormal cells, and the data space autocorrelation analysis is performed on the abnormal cells to obtain the missing data point set;

[0021] According to the absolute value of the time missing score of the abnormal cell, the missing data points of the abnormal cell are retained in whole or in part;

[0022] The missing values ​​in the multi-source safety standard dataset are marked according to the retention of missing data points in each abnormal cell.

[0023] On the basis of the above technical solution, preferably, the calculation formula of the time missing fraction is:

[0024]

[0025] in, is the time missing fraction of the ith cell, S i is the time missing statistics value of the i-th cell, For S i The variance of i is the number of data points in the ith cell, m i is the number of data sources for the ith cell, is the weight of the kth data source of the ith cell, f 1 (·) is the first time decay function, is the ath data point of the kth data source of the ith cell, is the bth data point of the kth data source of the ith cell, is the time difference between the a-th data point and the b-th data point of the k-th data source of the i-th cell, is the number of data points of the dth group of repeated values ​​in the ith cell, D is the total number of groups of repeated values ​​in the ith cell, τ 1 is the normalization factor, λ 1 is the attenuation coefficient of the first time attenuation function, and sgn(·) is the sign function.

[0026] On the basis of the above technical solution, preferably, the calculation formula of the region missing score is:

[0027]

[0028] Among them, Q qy (j) is the region missing fraction of the jth cell, f 2 (·) is the second time decay function, is the time difference between the central data point e of the jth cell and the central data point f of the adjacent cell, is the spatial weight between the central data point e of the jth cell and the adjacent central data point f, is the value of the adjacent central data point f, e j is the external data bias of the jth cell, is the sum of the spatial weights of the adjacent central data points f of the central data point e, is the mean of the multi-source security standard dataset, s is the standard deviation of the multi-source security standard dataset, n j is the number of data points in the jth cell, is the sum of the spatial weights of the adjacent central data points of the central data point e of the jth cell, λ 2 is the second time decay coefficient, f n is the number of adjacent central data points of the central data point e.

[0029] On the basis of the above technical solution, preferably, the data space autocorrelation analysis includes:

[0030]

[0031]

[0032]

[0033] in, is the missing outlier value of the g-th data point in the l-th outlier cell, s is the standard deviation of the multi-source security standard dataset, is the value of the g-th data point of the l-th abnormal cell, is the value of the hth data point of the lth abnormal cell, and the hth data point of the lth abnormal cell is the adjacent data point of the gth data point of the lth abnormal cell. is the mean of the multi-source security standard dataset, e l is the external data bias of the lth abnormal cell, is the spatial weight between the gth data point and the hth data point of the lth abnormal cell, h n is the number of adjacent data points of the g-th data point of the l-th abnormal cell, f 3 (·) is the third time decay function, is the time difference between the gth data point and the hth data point of the lth abnormal cell, D qs is the set of missing data points of the multi-source security standard dataset, is the subset of missing data points of the Nth abnormal cell, and N is the number of abnormal cells.

[0034] On the basis of the above technical solution, preferably, the missing data point retention status of each abnormal cell includes:

[0035]

[0036] in, is the missing data point retention of the I-th abnormal cell, is the time missing fraction of the I-th abnormal cell, are the missing score thresholds for the first, second, and third time periods, respectively.

[0037] On the basis of the above technical solution, preferably, the construction process of the region division model is:

[0038] Each area division scheme of the smart city is taken as a particle in the search space, and multiple particles are randomly generated according to the data distribution characteristics of the secure multi-dimensional data. Each particle includes the coordinates of K cluster centers.

[0039] The fitness value of each particle is calculated by the fitness function. The calculation formula of the fitness function is:

[0040] f syd =ω 1 G C +ω 2 G L +ω 3 G V +ω 4 G S ;

[0041] Among them, f syd is the fitness function, G C , G L , G V , G S are the regional management cost value, the inter-regional connection degree value, the internal difference value of the region, and the regional safety risk value, respectively. 1 ,ω 2 ,ω 3 ,ω 4 They are the weights of regional management cost value, inter-regional connection degree value, regional internal difference value, and regional security risk value;

[0042] The speed and position of the particle are updated through the speed update function and the position update function until the maximum number of iterations is reached to obtain the position of the global optimal particle. The global optimal particle corresponds to the regional division result of the smart city. The calculation formula of the speed update function and the position update function is:

[0043] v u,w (t l +1)=ω g v u,w (t l )+c 1r 1 (p u,w -x u,w (t l ))+c 2 r 2 (g w -x u,w (t l ));

[0044] x u,w (t l +1) = x u,w (t l )+v u,w (t l +1);

[0045] Among them, v u,w (t l +1) is the particle u in dimension w t l The speed at time +1, ω g is the inertia weight, v u,w (t l ) is the current velocity of particle u in dimension w, c 1 is the individual learning factor, c 2 is the social learning factor, r 1 and r 2 are the first and second random configurations, p u,w is the individual optimal position of particle u in dimension w, x u,w (t l ) is the current position of particle u in dimension w, g w is the global optimal position on dimension w, x u,w (t l +1) is the particle u at latitude w l +1 position at the moment.

[0046] On the basis of the above technical solutions, preferably, the inspection and management solution includes:

[0047] Evaluate the management cost of each area, reduce the inspection frequency for areas with high management cost values, and increase the inspection frequency for areas with low management cost values;

[0048] Analyze the degree of connection between regions, design multi-region patrol routes for regions with high regional connection values, and design single-region patrol routes for regions with low regional connection values;

[0049] Assess the internal differences of each area. For areas with high internal differences, divide them into sub-areas and adopt different inspection strategies for different sub-areas. For areas with low internal differences, unify the inspection strategy.

[0050] Analyze the safety risks of each area, increase the frequency of inspections for areas with high safety risk values, and conduct regular inspections for areas with low safety risk values.

[0051] The multi-dimensional data analysis inspection and management platform of the present invention has the following beneficial effects compared with the prior art:

[0052] (1) By collecting multi-source security data, pre-processing and multi-dimensionally analyzing the data, building a regional division model, and realizing the regional division of smart cities, a targeted inspection and governance plan is designed based on the regional division results, taking into account multiple factors such as regional management costs, regional connection, internal differences within regions, and security risks, thereby achieving more efficient and accurate management of smart cities;

[0053] (2) Using spatiotemporal data mining algorithms and data space autocorrelation analysis to identify and analyze multi-source security data sets can effectively detect data points that are spatially abnormal and improve the accuracy of data analysis;

[0054] (3) By introducing spatial weights and time decay functions, we can achieve comprehensive analysis of time and space dimensions, provide more comprehensive and accurate anomaly detection results, and enhance sensitivity to the latest trends;

[0055] (4) By introducing spatiotemporal data association rules and corresponding association rule functions, the correlation between data is comprehensively analyzed from the two dimensions of time and space, the accuracy of data interpolation is improved, and the interpolation results are more consistent with the actual distribution characteristics of the data;

[0056] (5) By calculating the scores of spatiotemporal data association rules and setting score thresholds, an adaptive rule selection mechanism is implemented to ensure that high-quality and highly relevant spatiotemporal data association rules are used for data interpolation, thereby improving the reliability of the interpolation results. The interpolation scores of each spatiotemporal data association rule are calculated, and the data correlation in both time and space dimensions is comprehensively considered to select the optimal interpolation rule, thereby balancing time continuity and spatial consistency and improving the overall quality of the interpolation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 This is a system structure diagram of a multi-dimensional data analysis inspection and management platform of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] See also Figure 1 This embodiment provides a patrol management platform for multi-dimensional data analysis, including:

[0061] A data collection unit, used to collect multi-source security data sets of smart cities;

[0062] A data processing unit, used to pre-process the multi-source security data set to obtain secure multi-dimensional data;

[0063] A data analysis unit is used to perform data analysis on the security multi-dimensional data to construct a regional division model, divide the smart city into regions based on the regional division model, and obtain a regional division result of the smart city, wherein the regional division result includes dividing the smart city into multiple regions, and each region has a corresponding regional management cost value, a regional connection degree value, a regional internal difference value, and a regional security risk value;

[0064] The inspection and management unit is used to design inspection and management plans based on the regional division results and manage the smart city based on the inspection and management plans.

[0065] Specifically, this embodiment collects multi-source security data, pre-processes and multi-dimensionally analyzes the data, constructs a regional division model, realizes the regional division of smart cities, and designs targeted patrol and management plans based on the regional division results, taking into account multiple factors such as regional management costs, regional connectivity, internal regional differences, and security risks, thereby achieving more efficient and accurate management of smart cities and solving data analysis and regional management problems in smart city patrol and management.

[0066] Specifically, the data collection unit is used to collect multi-source security data sets of the smart city;

[0067] The multi-source security data set includes geographic information data, environmental monitoring data, and video surveillance data.

[0068] Geographic information data includes urban area division information, road network and transportation facility distribution, location information of important places and facilities, and population density distribution data, which are used as the basis for calculating the subsequent regional management cost value and regional connection degree value;

[0069] Environmental monitoring data include air quality indicators, noise pollution levels, water quality monitoring data and meteorological data, which are used as the basis for calculating the subsequent regional internal difference values;

[0070] Video surveillance data includes traffic intersection surveillance videos, public place security surveillance videos, high-definition surveillance images of key areas, and dynamic video data collected by mobile cameras, which are used as the basis for calculating subsequent regional safety risk values.

[0071] The multi-source security data collection method of this embodiment provides comprehensive and real-time urban security information, and also provides an accurate data basis for subsequent regional division and security management, thereby improving the scientificity and effectiveness of smart city security management.

[0072] The pre-processing comprises:

[0073] The multi-source security data set is standardized to obtain a multi-source security standard data set, the multi-source security standard data set is identified through a spatiotemporal data mining algorithm to mark the missing values ​​in the multi-source security standard data set, the missing values ​​are interpolated based on the spatiotemporal data association rules to obtain a multi-source security complete data set, and multi-modal data fusion is performed on the multi-source security complete data set to obtain secure multi-dimensional data.

[0074] Specifically, the preprocessing of this embodiment performs Z-score standardization on the multi-source security data set to make the data from different sources at the same scale, eliminate the influence of the dimension, and obtain a multi-source security standard data set;

[0075] The multi-source safety standard data set is identified through spatiotemporal data mining algorithms, and the missing values ​​in the data set are marked, which helps to identify anomalies and missing parts in the data.

[0076] Data interpolation is performed on missing values ​​based on spatiotemporal data association rules to fill the gaps in the data set and obtain a multi-source secure and complete data set, thereby ensuring the integrity and continuity of the data.

[0077] Perform multimodal data fusion on multi-source security complete data sets, integrate different types of data (such as geographic information, environmental monitoring, video surveillance, etc.) into unified security multi-dimensional data, and improve the comprehensiveness of the data.

[0078] This embodiment eliminates the dimensional differences between different data sources through standardization processing, making data analysis more consistent and reliable. The spatiotemporal data mining algorithm can effectively identify anomalies and omissions in the data and ensure the accuracy of the data.

[0079] Through data interpolation, missing values ​​in the data set are filled, the integrity of the data is ensured, and a reliable data foundation is provided for subsequent analysis. Multimodal data fusion integrates different types of data into unified multi-dimensional data, enhancing the comprehensiveness and depth of data analysis.

[0080] Complete and standardized data sets improve the accuracy and reliability of analytical models, support more accurate regional division and management decisions, and through continuous mining and updating of spatiotemporal data, the platform can dynamically adjust data models to adapt to changes in the urban environment.

[0081] The standardization includes:

[0082] Perform Z-score normalization on the multi-source security data set to make the multi-source security data on the same scale:

[0083]

[0084] in, is the original data of the multi-source security standard dataset, is the original data of the multi-source security dataset, is the mean of the multi-source security dataset, is the standard deviation of the multi-source security dataset.

[0085] Specifically, this embodiment converts multi-source security data of different dimensions and ranges to the same scale through Z-score standardization. The mean of the standardized data is 0 and the standard deviation is 1, so that different types of data are comparable.

[0086] Identification of multi-source safety standard datasets through spatiotemporal data mining algorithms includes:

[0087] Arrange the multi-source security standard data set in time order and space order to obtain a multi-source security space-time volume, and divide the multi-source security space-time volume into a plurality of cells, each cell including an equal number of data points;

[0088] Perform time series trend analysis on each cell to obtain the time missing score of each cell;

[0089] Perform spatial abnormal area analysis on each cell to obtain the area missing score of each cell;

[0090] The cells corresponding to the absolute values ​​of the regional missing scores that are greater than the preset values ​​are divided into abnormal cells, and the data space autocorrelation analysis is performed on the abnormal cells to obtain the missing data point set;

[0091] According to the absolute value of the time missing score of the abnormal cell, the missing data points of the abnormal cell are retained in whole or in part;

[0092] The missing values ​​in the multi-source safety standard dataset are marked according to the retention of missing data points in each abnormal cell.

[0093] Specifically, this embodiment identifies and locates missing points and abnormal areas in a multi-source safety standard data set through time series trend analysis and spatial abnormal area analysis, thereby improving the integrity and reliability of the data.

[0094] By performing autocorrelation analysis on abnormal cells and intelligently screening and marking missing data points, the quality of the data set is improved, providing a more reliable data basis for subsequent analysis, thereby improving analysis accuracy.

[0095] Through multi-dimensional spatiotemporal data mining, more comprehensive and in-depth decision-making support information is provided to urban managers, which helps to optimize resource allocation and management strategies.

[0096] Through dynamic data management and exception handling, the platform's adaptability and stability in the face of complex and changeable urban data are enhanced, ensuring the reliable operation of the platform.

[0097] The calculation formula of the time missing fraction is:

[0098]

[0099] in, is the time missing fraction of the ith cell, S i is the time missing statistics value of the i-th cell, For S i The variance of i is the number of data points in the ith cell, m i is the number of data sources for the ith cell, is the weight of the kth data source of the ith cell, f 1 (·) is the first time decay function, is the ath data point of the kth data source of the ith cell, is the bth data point of the kth data source of the ith cell, is the time difference between the a-th data point and the b-th data point of the k-th data source of the i-th cell, is the number of data points of the dth group of repeated values ​​in the ith cell, D is the total number of groups of repeated values ​​in the ith cell, τ 1 is the normalization factor, λ 1 is the attenuation coefficient of the first time attenuation function, and sgn(·) is the sign function.

[0100] Specifically, the time missing score of this embodiment considers the data source of multi-source data, assigns a weight to each data source, and adjusts its impact on the overall trend according to the importance or reliability of different data sources;

[0101] By introducing a first time decay function, which is specifically an exponential decay function in this embodiment, considering that recent data may better reflect the current trend, more emphasis is placed on recent data points.

[0102] By calculating the weighted difference for each pair of data points, the comprehensive trend of multi-source data can be accurately captured.

[0103] By introducing data source weights in variance calculation, it is used to reflect the contribution of each data source to the overall trend uncertainty and improve the accuracy of statistical tests.

[0104] In this embodiment, the data source weight is introduced And multiple source data points In comparison, this method can effectively process data from different sources and perform weighted calculations based on the importance of each data source, making the analysis results more comprehensive and accurate and better reflecting the complex smart city environment.

[0105] By introducing the sign function and repeated value processing, this embodiment can effectively identify and process abnormal values ​​and repeated data, improve data quality, reduce the impact of noise on analysis results, and enhance the reliability of analysis.

[0106] By introducing multiple bias parameters, this embodiment provides a high degree of flexibility and adjustability, so that the platform can be optimized and adjusted according to specific application scenarios and requirements to adapt to different smart city management requirements.

[0107] In one embodiment, the normalization factor τ 1 The value is 18.

[0108] The calculation formula of the region missing score is:

[0109]

[0110] Among them, Q qy (j) is the region missing fraction of the jth cell, f 2 (·) is the second time decay function, is the time difference between the central data point e of the jth cell and the central data point f of the adjacent cell, is the spatial weight between the central data point e of the jth cell and the adjacent central data point f, is the value of the adjacent central data point f, e j is the external data bias of the jth cell, is the sum of the spatial weights of the adjacent central data points f of the central data point e, X is the mean of the multi-source security standard dataset, s is the standard deviation of the multi-source security standard dataset, and n j is the number of data points in the jth cell, is the sum of the spatial weights of the adjacent central data points of the central data point e of the jth cell, λ 2 is the second time decay coefficient, f n is the number of adjacent central data points of the central data point e.

[0111] Specifically, considering the change in the influence of time factors on data points, earlier data points may have less influence on the current analysis, and therefore need to be adjusted through a second time decay function. In this embodiment, an exponential decay function is specifically used.

[0112] Introducing external data bias j To integrate external information, including environmental factors and socio-economic data, and improve the accuracy of the analysis results, in this embodiment, each data point can be adjusted according to the degree of influence of external factors, and the value of external data bias can be determined according to the specific external data source and the degree of influence.

[0113] By introducing spatial weights To represent the spatial relationship between data points, ensure that spatially adjacent data points have greater influence in the analysis. In this embodiment, the spatial weight can be defined based on distance or proximity, such as binary (0 or 1) or distance-based weight.

[0114] In this embodiment, the second time decay function enables dynamic adjustment of the influence of data points, and the external data bias enables consideration of more external factors, thereby improving the ability to identify abnormal regions and data points.

[0115] This embodiment captures the spatial correlation of data by introducing spatial weights and the values ​​of adjacent data points, so that the analysis results can reflect the characteristics and anomalies of the data in spatial distribution, which helps to identify abnormal patterns or risk hotspots in local areas.

[0116] By combining the second time decay function and the spatial weight, this embodiment realizes the comprehensive analysis of the time and space dimensions. This integration enables the platform to consider both the time evolution and spatial distribution of data, providing a more comprehensive anomaly detection capability.

[0117] By introducing external data bias and global statistical mean And standard deviation s, this embodiment takes into account the global data distribution while considering the local features, so that the analysis result can reflect the local anomaly without ignoring the overall trend, thereby improving the accuracy and reliability of anomaly detection.

[0118] The data space autocorrelation analysis includes:

[0119]

[0120] in, is the missing outlier value of the g-th data point in the l-th outlier cell, s is the standard deviation of the multi-source security standard dataset, is the value of the g-th data point of the l-th abnormal cell, is the value of the hth data point of the lth abnormal cell, and the hth data point of the lth abnormal cell is the adjacent data point of the gth data point of the lth abnormal cell. is the mean of the multi-source security standard dataset, e l is the external data bias of the lth abnormal cell, is the spatial weight between the gth data point and the hth data point of the lth abnormal cell, h n is the number of adjacent data points of the g-th data point of the l-th abnormal cell, f 3 (·) is the third time decay function, is the time difference between the gth data point and the hth data point of the lth abnormal cell, D qs is the set of missing data points of the multi-source security standard dataset, is the subset of missing data points of the Nth abnormal cell, and N is the number of abnormal cells.

[0121] Specifically, in this embodiment, the time dimension of the data is taken into consideration so that newer data has a greater impact on the analysis results. Therefore, a third time decay function is introduced to assign different weights to data at different times. In this embodiment, a linear decay function is specifically used to better reflect the time dynamics of the data and improve sensitivity to the latest trends.

[0122] By calculating the missing outlier value of each data point, this embodiment can accurately identify local anomalies. This embodiment takes into account the relationship between a data point and its neighboring points, as well as the deviation from the global statistical characteristics, so that data points that are abnormal in space can be effectively detected.

[0123] By introducing spatial weights and the third time decay function, this embodiment achieves a comprehensive analysis of the time and space dimensions. This combination enables the platform to simultaneously consider the spatial correlation and time evolution characteristics of the data, providing more comprehensive and accurate anomaly detection results.

[0124] By constructing a missing data point set and a missing data point subset of an abnormal cell, this embodiment provides a systematic abnormal data management framework.

[0125] The missing data point retention conditions of each abnormal cell include:

[0126]

[0127] in, is the missing data point retention of the I-th abnormal cell, is the time missing fraction of the I-th abnormal cell, are the missing score thresholds for the first, second, and third time periods, respectively.

[0128] Specifically, by setting different time-missing score thresholds, this embodiment implements an adaptive data retention strategy. According to different degrees of time-missing scores, the system can flexibly decide to retain all or part of the missing data points, thereby achieving a balance between data integrity and data quality.

[0129] pass Each abnormal cell is evaluated and classified into four retention levels (100%, 80%, 60%, 20%). This embodiment provides a fine-grained data processing mechanism. This hierarchical processing can more accurately control data quality while retaining sufficient information for subsequent analysis.

[0130] In a specific embodiment, The values ​​can be 3.29, 2.58, and 1.96 respectively.

[0131] The data interpolation includes interpolating missing values ​​in a multi-source safety standard data set:

[0132]

[0133] in, They are the first spatiotemporal data association rule, the second spatiotemporal data association rule and the Mth spatiotemporal data association rule, are the first spatiotemporal data association rule function, the second spatiotemporal data association rule function, and the Mth spatiotemporal data association rule function, respectively. is the spatiotemporal data association value of the oth missing data point, and A'(o) is the interpolation value of the oth missing data point;

[0134] The calculation formula for the spatiotemporal data association rule score is:

[0135]

[0136] Among them, G α is the spatiotemporal data association rule score of the αth spatiotemporal data association rule, is the number of values ​​corresponding to the βth data of the αth spatiotemporal data association rule, is the number of data in the αth spatiotemporal data association rule, is the number of missing data in the αth spatiotemporal data association rule, R α is the number of non-missing data in the αth spatiotemporal data association rule, is the time missing fraction value of the γth cell, I α is the number of cells, Q qy (n) is the region missing score of the nth abnormal cell, and N is the number of abnormal cells;

[0137] Set the score threshold of spatiotemporal data association rules, and filter out association rules whose scores are greater than or equal to the score threshold as valid spatiotemporal data association rules:

[0138] Calculate the interpolation score of the spatiotemporal data association rule and select the spatiotemporal data association rule with the largest interpolation score for data interpolation:

[0139]

[0140] in, is the interpolation score of the αth spatiotemporal data association rule for the qth missing data point, A α (p) is the pth data of the αth spatiotemporal data association rule, f g (A α (q)) is the spatiotemporal data association function of the αth spatiotemporal data association rule for the qth missing data point and the time index, f t (A α (q)) is the spatiotemporal data association function of the αth spatiotemporal data association rule for the qth missing data point and the spatial index, n A is the number of data in the multi-source security standard dataset.

[0141] Specifically, this embodiment introduces multiple spatiotemporal data association rules and corresponding association rule functions to comprehensively analyze the association between data from two dimensions, time and space. This multi-dimensional analysis improves the accuracy of interpolation and makes the interpolation results more consistent with the actual distribution characteristics of the data.

[0142] By calculating the spatiotemporal data association rule scores and setting the score thresholds, this embodiment implements an adaptive rule selection mechanism, thereby ensuring that high-quality and highly relevant spatiotemporal data association rules are used for data interpolation, thereby improving the reliability of the interpolation results.

[0143] By calculating the interpolation score of each spatiotemporal data association rule, this embodiment comprehensively considers the data association in both time and space dimensions. This comprehensive evaluation mechanism can select the optimal interpolation rule, balance time continuity and spatial consistency, and improve the overall quality of the interpolation result.

[0144] This embodiment allows the definition of multiple spatiotemporal data association rules and association functions, which is highly flexible and scalable. This allows the platform to customize and optimize interpolation strategies according to different data characteristics and application scenarios, and adapt to the complex and changing smart city data environment.

[0145] Specifically, in a specific embodiment, the process of performing multimodal data fusion on the multi-source secure and complete data set after data interpolation is completed in this embodiment includes:

[0146] Divide the overall data into multiple Brodmann areas according to different functional areas or modules;

[0147] The generalized tensor discriminant analysis method is applied to the data of each region for dimensionality reduction to reduce the risk of overfitting;

[0148] Separate state estimation is performed on the dimensionally reduced data of each region using logistic tensor regression;

[0149] The tensor supervised decision-level fusion method is used to integrate the estimation results of each region to obtain safe multi-dimensional data.

[0150] The construction process of the region division model is as follows:

[0151] Each area division scheme of the smart city is taken as a particle in the search space, and multiple particles are randomly generated according to the data distribution characteristics of the secure multi-dimensional data. Each particle includes the coordinates of K cluster centers.

[0152] The fitness value of each particle is calculated by the fitness function. The calculation formula of the fitness function is:

[0153] f syd =ω 1 G C +ω 2 G L +ω 3 G V +ω 4 G S ;

[0154] Among them, f syd is the fitness function, G C , G L , G V , G S are the regional management cost value, the inter-regional connection degree value, the internal difference value of the region, and the regional safety risk value, respectively. 1 ,ω 2 ,ω 3 ,ω 4They are the weights of regional management cost value, inter-regional connection degree value, regional internal difference value, and regional security risk value;

[0155] The speed and position of the particle are updated through the speed update function and the position update function until the maximum number of iterations is reached to obtain the position of the global optimal particle. The global optimal particle corresponds to the regional division result of the smart city. The calculation formula of the speed update function and the position update function is:

[0156] v u,w (t l +1)=ω g v u,w (t l )+c 1 r 1 (p u,w -x u,w (t l ))+c 2 r 2 (g w -x u,w (t l ));

[0157] x u,w (t l +1) = x u,w (t l )+v u,w (t l +1);

[0158] Among them, v u,w (t l +1) is the particle u in dimension w t l The speed at time +1, ω g is the inertia weight, v u,w (t l ) is the current velocity of particle u in dimension w, c 1 is the individual learning factor, c 2 is the social learning factor, r 1 and r 2 are the first and second random configurations, p u,w is the individual optimal position of particle u in dimension w, x u,w (t l ) is the current position of particle u in dimension w, g w is the global optimal position on dimension w, x u,w (t l +1) is the particle u at latitude w l +1 position at the moment.

[0159] Specifically, this embodiment achieves multi-dimensional comprehensive optimization by introducing evaluation indicators in four dimensions: regional management cost, degree of inter-regional connection, internal regional differences, and regional security risks. By comprehensively considering all aspects of smart city management, it ensures that the division results are balanced in multiple key indicators, thereby improving the overall efficiency and effectiveness of urban management.

[0160] The particle swarm optimization algorithm is used to continuously adjust the position of particles through the speed update function and the position update function to realize an adaptive search strategy. This embodiment can effectively avoid the local optimal solution and improve the search efficiency of the global optimal solution, thereby finding a better regional division scheme in a complex urban environment.

[0161] By setting the maximum number of iterations and updating the velocity and position of the particles in each iteration, the region partitioning model achieves a continuous optimization process. This iterative mechanism can continuously improve the region partitioning scheme and gradually approach the global optimal solution, thereby improving the quality and reliability of the final partitioning result.

[0162] The inspection and governance plan includes:

[0163] Evaluate the management cost of each area, reduce the inspection frequency for areas with high management cost values, and increase the inspection frequency for areas with low management cost values;

[0164] Analyze the degree of connection between regions, design multi-region patrol routes for regions with high regional connection values, and design single-region patrol routes for regions with low regional connection values;

[0165] Assess the internal differences of each area. For areas with high internal differences, divide them into sub-areas and adopt different inspection strategies for different sub-areas. For areas with low internal differences, unify the inspection strategy.

[0166] Analyze the safety risks of each area, increase the frequency of inspections for areas with high safety risk values, and conduct regular inspections for areas with low safety risk values.

[0167] Specifically, this embodiment further refines the inspection and management through the attributes of each area, including the area management cost value, the area connection degree value, the area internal difference value and the area safety risk value.

[0168] Optimize patrol resource allocation according to regional attributes and improve resource utilization efficiency.

[0169] According to the management cost value, patrol personnel and equipment are rationally allocated to ensure efficient use.

[0170] Use regional connectivity values ​​to optimize patrol team collaboration and information sharing.

[0171] For areas with high internal differences, a variety of inspection tools and methods are configured.

[0172] This embodiment includes dynamic patrol strategy adjustment, which dynamically adjusts the patrol strategy according to real-time data and area attributes.

[0173] Monitor regional safety risk values ​​in real time and adjust patrol frequency and routes in a timely manner.

[0174] Adjust the inspection team's collaboration model according to changes in regional connectivity.

[0175] Regularly evaluate internal differences within the region and optimize the allocation of inspection tasks.

[0176] This embodiment also includes comprehensive management and decision support, using regional attribute data to support urban management decisions and improve management levels.

[0177] Utilize management cost value data to optimize city budget allocation and improve the efficiency of fund use.

[0178] Support urban planning and infrastructure construction through regional connectivity analysis.

[0179] Formulate differentiated urban management policies based on internal regional differences and security risk values.

[0180] This embodiment also includes a public participation and feedback mechanism to enhance public participation and improve the transparency and responsiveness of urban management.

[0181] Develop a public participation platform to collect citizens’ feedback and suggestions on regional management.

[0182] Dynamically adjust regional management strategies and patrol plans based on citizen feedback.

[0183] Regularly publish regional management reports to enhance public trust in urban management.

[0184] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A patrol management platform for multi-dimensional data analysis, characterized in that: include: A data collection unit, used to collect multi-source security data sets of smart cities; A data processing unit, used to pre-process the multi-source security data set to obtain secure multi-dimensional data; A data analysis unit is used to perform data analysis on the security multi-dimensional data to construct a regional division model, divide the smart city into regions based on the regional division model, and obtain a regional division result of the smart city, wherein the regional division result includes dividing the smart city into multiple regions, and each region has a corresponding regional management cost value, a regional connection degree value, a regional internal difference value, and a regional security risk value; The inspection and management unit is used to design inspection and management plans based on the regional division results and manage the smart city based on the inspection and management plans.

2. A multi-dimensional data analysis inspection and management platform as claimed in claim 1, characterized in that: The pre-processing comprises: The multi-source security data set is standardized to obtain a multi-source security standard data set, the multi-source security standard data set is identified through a spatiotemporal data mining algorithm to mark the missing values ​​in the multi-source security standard data set, the missing values ​​are interpolated based on the spatiotemporal data association rules to obtain a multi-source security complete data set, and multi-modal data fusion is performed on the multi-source security complete data set to obtain secure multi-dimensional data.

3. A multi-dimensional data analysis inspection and management platform as claimed in claim 2, characterized in that: The standardization includes: Perform Z-score normalization on the multi-source security data set to make the multi-source security data on the same scale: in, is the original data of the multi-source security standard dataset, is the original data of the multi-source security dataset, is the mean of the multi-source security dataset, is the standard deviation of the multi-source security dataset.

4. A multi-dimensional data analysis inspection and management platform as claimed in claim 3, characterized in that: Identification of multi-source safety standard datasets through spatiotemporal data mining algorithms includes: Arrange the multi-source security standard data set in time order and space order to obtain a multi-source security space-time volume, and divide the multi-source security space-time volume into a plurality of cells, each cell including an equal number of data points; Perform time series trend analysis on each cell to obtain the time missing score of each cell; Perform spatial abnormal area analysis on each cell to obtain the area missing score of each cell; The cells corresponding to the absolute values ​​of the regional missing scores that are greater than the preset values ​​are divided into abnormal cells, and the data space autocorrelation analysis is performed on the abnormal cells to obtain the missing data point set; According to the absolute value of the time missing score of the abnormal cell, the missing data points of the abnormal cell are retained in whole or in part; The missing values ​​in the multi-source safety standard dataset are marked according to the retention of missing data points in each abnormal cell.

5. A multi-dimensional data analysis inspection and management platform as claimed in claim 4, characterized in that: The calculation formula of the time missing fraction is: in, is the time missing fraction of the ith cell, S i is the time missing statistics value of the i-th cell, For S i The variance of i is the number of data points in the ith cell, m i is the number of data sources for the ith cell, is the weight of the kth data source of the ith cell, f1(·) is the first time decay function, is the ath data point of the kth data source of the ith cell, is the bth data point of the kth data source of the ith cell, is the time difference between the a-th data point and the b-th data point of the k-th data source of the i-th cell, is the number of data points of the dth group of repeated values ​​in the ith cell, D is the total number of groups of repeated values ​​in the ith cell, τ1 is the normalization factor, λ1 is the decay coefficient of the first time decay function, and sgn(·) is the sign function.

6. A multi-dimensional data analysis inspection and management platform as claimed in claim 5, characterized in that: The calculation formula of the region missing score is: Among them, Q qy (j) is the region missing fraction of the jth cell, f2(·) is the second time decay function, is the time difference between the central data point e of the jth cell and the central data point f of the adjacent cell, is the spatial weight between the central data point e of the jth cell and the adjacent central data point f, is the value of the adjacent central data point f, e j is the external data bias of the jth cell, is the sum of the spatial weights of the adjacent central data points f of the central data point e, is the mean of the multi-source security standard dataset, s is the standard deviation of the multi-source security standard dataset, n j is the number of data points in the jth cell, is the spatial weight square sum of the neighboring central data points of the central data point e of the jth cell, λ2 is the second time decay coefficient, and f n is the number of adjacent central data points of the central data point e.

7. A multi-dimensional data analysis inspection and management platform as claimed in claim 6, characterized in that: The data space autocorrelation analysis includes: in, is the missing outlier value of the g-th data point in the l-th outlier cell, s is the standard deviation of the multi-source security standard dataset, is the value of the g-th data point of the l-th abnormal cell, is the value of the hth data point of the lth abnormal cell, and the hth data point of the lth abnormal cell is the adjacent data point of the gth data point of the lth abnormal cell. is the mean of the multi-source security standard dataset, e l is the external data bias of the lth abnormal cell, is the spatial weight between the gth data point and the hth data point of the lth abnormal cell, h n is the number of adjacent data points of the g-th data point of the l-th abnormal cell, f3(·) is the third time decay function, is the time difference between the gth data point and the hth data point of the lth abnormal cell, D qs is the set of missing data points of the multi-source security standard dataset, is the subset of missing data points of the Nth abnormal cell, and N is the number of abnormal cells.

8. A multi-dimensional data analysis inspection and management platform as claimed in claim 7, characterized in that: The missing data point retention conditions of each abnormal cell include: in, is the missing data point retention of the I-th abnormal cell, is the time missing fraction of the I-th abnormal cell, are the missing score thresholds for the first, second, and third time periods, respectively.

9. The multi-dimensional data analysis inspection and management platform as claimed in claim 1, characterized in that: The construction process of the region division model is as follows: Each area division scheme of the smart city is taken as a particle in the search space, and multiple particles are randomly generated according to the data distribution characteristics of the secure multi-dimensional data. Each particle includes the coordinates of K cluster centers. The fitness value of each particle is calculated by the fitness function. The calculation formula of the fitness function is: f syd =ω1G C +ω2G L +ω3G V +ω4G S ; Among them, f syd is the fitness function, G C , G L , G V , G S are the regional management cost value, the inter-regional connection degree value, the regional internal difference value, and the regional safety risk value, respectively; ω1, ω2, ω3, and ω4 are the weights of the regional management cost value, the inter-regional connection degree value, the regional internal difference value, and the regional safety risk value, respectively; The speed and position of the particle are updated through the speed update function and the position update function until the maximum number of iterations is reached to obtain the position of the global optimal particle. The global optimal particle corresponds to the regional division result of the smart city. The calculation formula of the speed update function and the position update function is: v u,w (t l +1)=ω g v u,w (t l )+c1r1(p u,w -x u,w (t l ))+c2r2(g w -x u,w (t l )); x u,w (t l +1)=x u,w (t l )+v u,w (t l +1); Among them, v u,w (t l +1) is the particle u in dimension w t l The speed at time +1, ω g is the inertia weight, v u,w (t l ) is the current velocity of particle u in dimension w, c1 is the individual learning factor, c2 is the social learning factor, r1 and r2 are the first and second random configurations, p u,w is the individual optimal position of particle u in dimension w, x u,w (t l ) is the current position of particle u in dimension w, g w is the global optimal position on dimension w, x u,w (t l +1) is the particle u at latitude w l +1 position at the moment.

10. A multi-dimensional data analysis inspection and management platform as claimed in claim 9, characterized in that: The inspection and governance plan includes: Evaluate the management cost of each area, reduce the inspection frequency for areas with high management cost values, and increase the inspection frequency for areas with low management cost values; Analyze the degree of connection between regions, design multi-region patrol routes for regions with high regional connection values, and design single-region patrol routes for regions with low regional connection values; Assess the internal differences of each area. For areas with high internal differences, divide them into sub-areas and adopt different inspection strategies for different sub-areas. For areas with low internal differences, unify the inspection strategy. Analyze the safety risks of each area, increase the frequency of inspections for areas with high safety risk values, and conduct regular inspections for areas with low safety risk values.

Images