Digitalization method and system of urban area model based on multi-source heterogeneous information collation

Abstract

The present application relates to the field of data processing technology, and specifically to a method and system for digitizing an urban area model based on multi-source heterogeneous information collation, the method comprising: collecting various types of data at various collection points in various fields in an urban area at various times; composing all types of data at various collection points at various times in various fields into feature vectors at various times; determining the initial abnormal value of the feature vector of each collection point at the current time in each field; obtaining the reference reliability of the feature vector of any collection point at the current time in each field and the remaining collection points; obtaining the first abnormal value of the feature vector of each collection point at the current time in each field; determining the second abnormal value of the feature vector of each collection point at the current time in each field; correcting the initial abnormal value, obtaining the processing results of all types of data of each collection point at the current time in each field, and digitizing. Thus, the accuracy of digitizing the urban area model is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for digitizing an urban area model based on the collation of multi-source heterogeneous information. Background Art

[0002] Urban model digitization refers to the process of converting actual information and data of urban areas into digital models. The collected heterogeneous data are integrated and processed to form a unified digital model, which usually involves data cleaning and denoising, ensuring that data from different sources can be seamlessly combined while improving the accuracy of digital model creation.

[0003] When cleaning and denoising multi-source heterogeneous data, it is necessary to detect outliers on the collected multi-source heterogeneous data. Usually, the existing anomaly detection algorithms fail to take into account the mutual influence between data from different collection points and the temporal correlation between data collected from a single collection point during the process of anomaly detection, resulting in low accuracy of data anomaly detection, which in turn affects the accuracy of urban model digitization. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a method and system for digitizing urban area models based on multi-source heterogeneous information collation. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for digitizing an urban area model based on multi-source heterogeneous information collation, the method comprising the following steps:

[0006] Collect all types of data at all collection points in all fields in the urban area at all times; combine all types of data at all collection points in all fields at all times into feature vectors at all times;

[0007] The feature vectors of all historical moments of each collection point at the current moment in each field are divided into clusters, and the initial abnormal value of the feature vector of each collection point at the current moment in each field is determined based on the difference in the degree of membership of the feature vector of each collection point at the current moment in each field in different clusters; the correlation between any two types of data at the current moment and all historical moments of each collection point in each field is analyzed, and the reference reliability of the feature vector of the current moment of any collection point and the remaining collection points in each field is determined based on the difference in the correlation between any collection point and the remaining collection points in each field, and the difference between the same type of data of any collection point and the remaining collection points;

[0008] Divide the various types of data at the current moment and all historical moments of each collection point in each field into various data clusters, and determine the abnormality degree of any type of data at the current moment of each collection point in each field based on the difference between the data cluster where any type of data is located at the current moment and the remaining data clusters of the any type of data; determine the first abnormal value of the feature vector of each collection point in each field at the current moment based on the abnormality degree of all types of data at the current moment of each collection point in each field;

[0009] Based on all the reference reliabilities of any collection point in each field and the differences between the same type of data, the second outlier value of the feature vector of each collection point in each field at the current moment is determined; the initial outlier value is corrected by combining the first outlier value and the second outlier value to obtain the corrected outlier value of the feature vector of each collection point in each field at the current moment, and the processing results of all types of data of each collection point in each field at the current moment are obtained and digitized.

[0010] In one embodiment, the process of determining the initial outlier is:

[0011] For any feature vector of each collection point at the current moment, the membership of any feature vector in each cluster cluster is counted, the absolute value of the difference between the membership of any feature vector in the cluster cluster where it is located and the membership of the remaining cluster clusters is calculated, and the normalized value of the sum of all the absolute values ​​of the difference of any feature vector is used as the initial outlier value of any feature vector.

[0012] In one embodiment, the determination of the reference reliability includes:

[0013] The current moment of each collection point in each field and all types of data at all historical moments are combined into each data sequence at the current moment, and the correlation between any two data sequences at the current moment of each collection point in each field is calculated, recorded as the first correlation, and the differences between all the first correlations of any collection point and the remaining collection points are analyzed to determine the parameter distribution similarity of the feature vectors at the current moment of any collection point in each field and the remaining collection points;

[0014] Calculate the correlation between the data sequence of the same type at the current moment of any collection point in each field and the remaining collection points, record it as the second correlation, and take the normalized value of the sum of all the second correlations at the current moment of any collection point in each field and the remaining collection points as the parameter value similarity of the feature vector at the current moment of any collection point in each field and the remaining collection points;

[0015] The product of the parameter distribution similarity and the parameter value similarity is used as the reference reliability.

[0016] In one embodiment, the parameter distribution similarity is calculated as follows:

[0017] ; For collection point With collection point The similarity of the parameter distribution of the feature vector A at the current moment, For collection point With collection point of Type data, the absolute value of the difference of the first correlation between the types of data, is the number of types contained in the eigenvector A, and exp() represents the exponential function with a natural constant as the base.

[0018] In one embodiment, the process of determining the abnormality degree is:

[0019] Taking multiple elements connected in time series in each data cluster as a sequence segment, if the data of any type at each acquisition point at the current moment in each field is located in the sequence segment, calculating the absolute value of the difference between the number of sequence segments in the data cluster where the data of any type is located and the number of sequence segments in the remaining data clusters, recording it as the first absolute value of the difference, calculating the sum of all the first absolute values ​​of the difference of the data of any type, recording it as the first sum, calculating the metric distance between the sequence segment where the data of any type is located and the remaining sequence segments, calculating the sum of all the metric distances of the data of any type, recording it as the second sum;

[0020] The product of the first sum value and the second sum value is used as an abnormal factor of any type of data at the current moment;

[0021] Determine the abnormal significance of any type of data at the current moment based on the abnormal factor;

[0022] The normalized value of the product of the abnormal factor and the abnormal significance is used as the abnormal degree. If any type of data is not located in the sequence segment, the abnormal degree is set to a preset value.

[0023] In one embodiment, the abnormal significance is calculated as follows:

[0024] ; is the abnormal significance of any type of data at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing the o-th data in the sequence segment where any type of data is located at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing any type of data from the sequence segment where any type of data is located at the current moment, and D is the number of data contained in the sequence segment where any type of data is located at the current moment.

[0025] In one embodiment, the first abnormal value is a normalized value of the sum of the abnormality levels of all types of data at each collection point in each field at the current moment.

[0026] In one embodiment, the second outlier value is calculated as follows:

[0027] ; For collection point The second outlier value of the feature vector A at the current moment, For collection point With collection point The reference reliability of the feature vector A at the current moment, M is the number of acquisition points, For collection point With collection point The absolute value of the difference between the h type data at the current moment, is the number of types contained in the feature vector A, and norm() is the normalization function.

[0028] In one embodiment, the processing result of obtaining all types of data at each collection point in each field at the current moment includes:

[0029] Calculate the mean of the first outlier value and the second outlier value, the corrected outlier value is the product of the initial outlier value and the mean, if the corrected outlier value is greater than or equal to the preset outlier threshold, then remove all types of data in the feature vector corresponding to the current moment, otherwise, retain all types of data in the feature vector corresponding to the current moment.

[0030] In a second aspect, an embodiment of the present application further provides a digital urban model system based on the organization of multi-source heterogeneous information, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned methods when executing the computer program.

[0031] This application has at least the following beneficial effects:

[0032] The present application collects various types of data at various collection points in various fields in the urban area at various times; composes all types of data at various times of various collection points in various fields into feature vectors at various times; divides the feature vectors at all historical times of the current time of various collection points in various fields into various clusters, and determines the initial abnormal value of the feature vector of the current time of various collection points in various fields based on the difference in the membership of the feature vector of the current time of various collection points in various fields in different clusters, so as to make an initial judgment on the abnormal situation of all types of data at the current time of various collection points in various fields; further, analyzes the current time of various collection points in various fields and their The reference reliability of the feature vectors of the current moment of any collection point and the other collection points in each field is determined based on the correlation between any two types of data at historical moments. The reference reliability reflects the correlation between the collected data of each collection point and other collection points, and improves the reliability of abnormal judgment on the collected data of each collection point. Furthermore, the current moment of each collection point in each field and all types of data at historical moments are divided into data clusters. The difference between the data cluster where any type of data at the moment is located and the remaining data cluster of any type of data is determined to determine the abnormal degree of any type of data at the current moment of each collection point in each field; based on the abnormal degree of all types of data at the current moment of each collection point in each field, the first abnormal value of the feature vector of each collection point at the current moment in each field is determined, and the first abnormal value reflects the temporal correlation between all data at the current moment and all historical moments of each type of data in each field of each collection point, thereby improving the accuracy of abnormal judgment on the collected data of each collection point; finally, based on all the reference reliability of any collection point in each field and the difference between the same type of data, the second abnormal value of the feature vector of each collection point at the current moment in each field is determined; combined with the first abnormal value and the second abnormal value, the initial abnormal value is corrected to obtain the corrected abnormal value of the feature vector of each collection point at the current moment in each field, the corrected abnormal value corrects the error of the initial abnormal value, improves the accuracy of determining the abnormal value of the feature vector, improves the accuracy of abnormal detection of the collected data, obtains the processing results of all types of data at the current moment of each collection point in each field, and digitizes them, thereby improving the accuracy of digitization of the urban model. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 A flowchart of the steps of a method for digitizing an urban area model based on multi-source heterogeneous information collation provided in one embodiment of the present application;

[0035] Figure 2 Flowchart for initial outlier correction. DETAILED DESCRIPTION

[0036] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation methods, structures, features and effects of the urban area model digitization method and system based on multi-source heterogeneous information collation proposed in the present application. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0037] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0038] The specific scheme of the urban area model digitization method and system based on multi-source heterogeneous information collation provided by the present application is described in detail below with reference to the accompanying drawings.

[0039] See also Figure 1 , which shows a flowchart of a method for digitizing an urban area model based on multi-source heterogeneous information collation provided by an embodiment of the present application, the method comprising the following steps:

[0040] S1, collect all types of data at all collection points in all fields in the urban area at all times; combine all types of data at all collection points in all fields at all times into feature vectors at all times.

[0041] This embodiment evenly sets various collection points in the urban area, and collects various types of data at various times in various fields at various collection points. The various fields include: urban geographic information field, road network field, infrastructure field, environmental monitoring field, power monitoring field, etc. The implementer can determine the various fields according to the actual situation, and this embodiment does not limit it here.

[0042] Each field contains various types of data. Taking the environmental monitoring field and the power monitoring field as examples, the environmental monitoring field includes: temperature, humidity, wind speed, and the power monitoring field includes: voltage, current, and power consumption.

[0043] In this embodiment, the collection points are shops in the urban area. The number of collection points in this embodiment is 100, and the collection frequency is once every 20 minutes. The implementer can set it according to the actual situation, and this embodiment does not limit it here.

[0044] It should be noted that the settings of the collection points and the types of data collected can be set by the implementer according to the actual situation, and this embodiment does not limit this. All the collected data are collectively referred to as multi-source heterogeneous data.

[0045] All types of data in each field are combined into feature vectors of the corresponding field at each moment. Taking the environmental monitoring field as an example, the temperature, humidity, and wind speed collected at each moment are combined into feature vectors of the environmental monitoring field at each moment.

[0046] S2, dividing the feature vectors of all historical moments of each collection point at the current moment in each field into clusters, and determining the initial abnormal value of the feature vector of each collection point at the current moment in each field based on the difference in the degree of membership of the feature vector of each collection point at the current moment in each field in different clusters; analyzing the correlation between any two types of data at the current moment and all historical moments of each collection point in each field, and determining the reference reliability of the feature vector of the current moment of any collection point and the remaining collection points in each field based on the difference in the correlation between any collection point and the remaining collection points in each field, and the difference between the same type of data of any collection point and the remaining collection points.

[0047] This embodiment uses environmental monitoring data as an example for analysis, and uses the feature vectors at the current moment and all its historical moments as the input of the FCM (Fuzzy C-means) clustering algorithm, and outputs each cluster and the membership of each feature vector to each cluster. Based on the difference in the membership of the feature vector at each moment in different clusters, the initial outlier value of the feature vector at each collection point at each moment is determined. The specific calculation method is:

[0048] ; In the formula, For collection point The initial outlier value of the feature vector A at the current moment, norm() is the normalization function, is the absolute value of the difference in membership between the cluster where the feature vector is located at the current moment and the w-th cluster, and K represents the number of clusters.

[0049] The FCM clustering algorithm is a well-known technology and will not be described in detail in this embodiment.

[0050] This embodiment takes the environmental monitoring field as an example for analysis, and the temperature data at each moment and all its historical moments form a temperature sequence at each moment, the humidity data at each moment and all its historical moments form a humidity sequence at each moment, and the wind speed data at each moment and all its historical moments form a wind speed sequence at each moment.

[0051] Calculate the correlation between any two types of data sequences in the environmental monitoring field at each collection point at each time, which is recorded as the first correlation. That is, in this embodiment, the correlation between the humidity sequence and the temperature sequence, the correlation between the humidity sequence and the wind speed sequence, and the correlation between the temperature sequence and the wind speed sequence are calculated respectively;

[0052] It should be noted that the correlation can be specifically calculated by using a Pearson correlation coefficient, a cosine similarity, or the like. In this embodiment, the Pearson correlation coefficient is used as a calculation method for the correlation.

[0053] Based on the correlation between any two types of data sequences in the environmental monitoring field at each collection point at each time, the parameter distribution similarity of the feature vectors of each collection point and the other collection points at each time is determined. The specific calculation method is:

[0054] ; For collection point With collection point The similarity of the parameter distribution of the feature vector A at the current moment, For collection point With collection point of Type data, the absolute value of the difference of the first correlation between the types of data, is the number of types contained in the eigenvector A, and exp() represents the exponential function with a natural constant as the base.

[0055] It should be understood that, when the difference in the first correlation between two acquisition points is greater, it means that the difference in the data collected by the two acquisition points is greater, and thus the similarity of the data between the two acquisition points is lower, that is, the similarity of the parameter distribution is smaller.

[0056] Further, the correlation between the same type of data of each collection point and the rest of the collection points is analyzed, specifically, the correlation between the data sequence of the same type at the current moment of any collection point in each field and the rest of the collection points is calculated, recorded as the second correlation, and the similarity of the parameter values ​​of the feature vectors of each collection point and the rest of the collection points at each moment is determined. The specific calculation method is:

[0057] ; In the formula, For collection point With collection point The similarity of the parameter value of the feature vector A at the current moment, norm() is the normalization function, Collect points for the current moment With collection point The second correlation between h types of data, is the number of types contained in the feature vector A.

[0058] Combine the similarity of the parameter distribution and the similarity of the parameter value to calculate the acquisition point With collection point The reference reliability of the feature vector A at the current moment is calculated as follows:

[0059] ; In the formula, For collection point With collection point The reference reliability of the feature vector A at the current moment.

[0060] It should be understood that, if the parameter distribution similarity between two collection points is greater, the parameter value similarity is greater, which means that the data in the corresponding fields between the two collection points are more mutually referenceable, that is, the reference reliability is greater.

[0061] For the feature vectors of each collection point in each field at the current moment, the same calculation method as the reference reliability is used to obtain the reference reliability of the feature vectors of each collection point and the other collection points in each field at the current moment.

[0062] S3, dividing the various types of data at the current moment and all historical moments of each collection point in each field into various data clusters, and determining the abnormality degree of any type of data at the current moment of each collection point in each field based on the difference between the data cluster where any type of data is located at the current moment and the remaining data clusters of the any type of data; and determining the first abnormal value of the feature vector of each collection point in each field at the current moment based on the abnormality degree of all types of data at the current moment of each collection point in each field.

[0063] The various types of data collected at each collection point at each moment and at all its historical moments are combined to form a data sequence of various types at each moment. Taking the temperature data in the field of environmental monitoring as an example, the temperature data at the current moment and all its previous historical moments of each collection point are combined to form a temperature sequence at the current moment. The temperature sequence at the current moment is used as the input of the DBSCAN clustering algorithm and output as each cluster cluster. Each cluster cluster output by the DBSCAN clustering algorithm is recorded as each data cluster. The elements that are continuous in time sequence in each data cluster are taken as a sequence segment to obtain each sequence segment of the temperature sequence at the current moment. For example, there is a data cluster with elements (1, 3, 2, 4, 5, 8), where 3 and 2 are continuous in time sequence, and 5 and 8 are continuous in time sequence. Then (3, 2) is taken as a sequence segment, and (5, 8) is taken as a sequence segment. Among them, the DBSCAN clustering algorithm is an existing well-known technology, and this embodiment will not be described in detail here, and the implementer can select other existing clustering algorithms according to actual conditions.

[0064] Similarly, taking the temperature data of any collection point as an example for analysis, if the temperature data of each collection point at the current moment is within the sequence segment, the abnormality of the temperature data of each collection point at the current moment is analyzed. The specific calculation method is:

[0065] ; Indicates the abnormal factor of the temperature data at each collection point at the current moment, It represents the absolute value of the difference between the number of sequence segments in the data cluster where the temperature data of each acquisition point is located at the current moment and the number of sequence segments in the rth data cluster. H represents the number of data clusters divided by the temperature sequence of each acquisition point at the current moment. is the DTW distance between the sequence segment where the temperature data of each acquisition point at the current moment is located and the qth sequence segment, and F represents the number of sequence segments into which the temperature sequence of each acquisition point at the current moment is divided.

[0066] It should be understood that if the number of sequence segments in the data cluster where the temperature data is located at the current moment is different from the number of sequence segments in other clusters, that is, The larger the value is, the more inconsistent the data distribution is between the data cluster where the current temperature data is located and other data clusters, the more likely the data in the data cluster where the current temperature data is located is to be abnormal, and the greater the abnormality of the current temperature data is; if the metric distance between the sequence segment where the current temperature data is located and the other sequence segments is larger, that is The larger it is, the greater the difference between the sequence segment where the current temperature data is located and the other sequence segments, the more likely it is that there is an anomaly in the sequence segment where the current temperature data is located, and the greater the anomaly factor of the current temperature data.

[0067] The same calculation method as the abnormal factor of the temperature data of each collection point at the current moment is used to calculate the abnormal factor of each type of data at each collection point at the current moment in each field. Taking any type of data at each collection point at the current moment in each field as an example for analysis, the abnormal significance of any type of data is calculated. The specific calculation method is:

[0068] ; is the abnormal significance of any type of data at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing the o-th data in the sequence segment where any type of data is located at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing any type of data from the sequence segment where any type of data is located at the current moment, and D is the number of data contained in the sequence segment where any type of data is located at the current moment.

[0069] The normalized value of the product of the abnormal factor and the abnormal significance of any type of data at the current moment is used as the abnormal degree of any type of data at the current moment. In this embodiment, the method for obtaining the normalized value is to use the Sigmoid function. The implementer can choose other existing normalization methods according to actual conditions, and this embodiment does not limit it here.

[0070] It should be noted that if any type of data at each collection point at the current moment is not within the sequence segment, the abnormality level of any type of data at the current moment is recorded as a preset value 1, which can be set by the implementer according to actual conditions, and this embodiment does not limit this.

[0071] Taking the current environmental monitoring field as an example, the first abnormal value of the feature vector A of each collection point at the current moment is calculated. The specific calculation method is:

[0072] ; In the formula, For collection point The first outlier value of the feature vector A at the current moment, For collection point The abnormality level of h type data at the current moment, is the number of types contained in the feature vector A, and norm() is the normalization function.

[0073] It should be understood that if the degree of abnormality of each type of data at each collection point in each field at the current moment is greater, and the first abnormal value of the corresponding feature vector in the field at the current moment is greater, it means that the possibility of abnormality in the corresponding feature vector is greater.

[0074] S4, based on all the reference reliabilities of any collection point in each field and the differences between the same type of data, determine the second abnormal value of the feature vector of each collection point in each field at the current moment; combine the first abnormal value and the second abnormal value to correct the initial abnormal value to obtain the corrected abnormal value of the feature vector of each collection point in each field at the current moment, obtain the processing results of all types of data of each collection point in each field at the current moment, and digitize them.

[0075] Calculate the second outlier value of the feature vector A of each collection point at the current moment. The specific calculation method is:

[0076] ; For collection point The second outlier value of the feature vector A at the current moment, For collection point With collection point The reference reliability of the feature vector A at the current moment, M is the number of acquisition points, For collection point With collection point The absolute value of the difference between the h type data at the current moment, is the number of types contained in the feature vector A, and norm() is the normalization function.

[0077] Combined with the first abnormal value and the second abnormal value of the feature vector A of each collection point at the current moment, the abnormal reference degree of the feature vector A of each collection point at the current moment is obtained, specifically: the average of the first abnormal value and the second abnormal value of the feature vector A of each collection point at the current moment is used as the abnormal reference degree of the feature vector A of each collection point at the current moment.

[0078] Based on the abnormal reference degree of the feature vector A at each acquisition point at the current moment, the initial abnormal value of the feature vector A is corrected, and the corrected abnormal value of the feature vector A at each acquisition point at the current moment is calculated. The specific calculation method is:

[0079] ; For collection point The corrected outlier value of the feature vector A at the current moment, For collection point The initial outlier value of the feature vector A at the current moment, For collection point The abnormal reference degree of the feature vector A at the current moment. The initial abnormal value correction flow chart is as follows Figure 2 shown.

[0080] Adoption and collection points The same calculation method of the corrected outlier value of the feature vector A at the current moment can be used to calculate the corrected outlier value of each feature vector at each acquisition point at the current moment.

[0081] An abnormal threshold is set. If the corrected abnormal value of the feature vector at the current moment is greater than or equal to the abnormal threshold, all types of data corresponding to the feature vector at the current moment are eliminated. Otherwise, all types of data contained in the feature vector at the current moment are retained. In this embodiment, the abnormal threshold is artificially set to 0.8, and the implementer can set it according to the actual situation. This embodiment does not limit it here.

[0082] At this point, the processing results of the current data of each collection point in each field can be obtained. The data collected at each moment is processed using the same data processing steps as the current moment to obtain the data processing results at each moment. The processed data is used for the digital creation of the urban model.

[0083] Based on the same inventive concept as the above method, an embodiment of the present application also provides a system for digitizing an urban model based on multi-source heterogeneous information collation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of any one of the above methods for digitizing an urban model based on multi-source heterogeneous information collation are implemented.

[0084] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0085] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0086] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for digitizing urban area models based on the collation of multi-source heterogeneous information, characterized in that: The method comprises the following steps: Collect all types of data at all collection points in all fields in the urban area at all times; combine all types of data at all collection points in all fields at all times into feature vectors at all times; The feature vectors of all historical moments of each collection point at the current moment in each field are divided into clusters, and the initial abnormal value of the feature vector of each collection point at the current moment in each field is determined based on the difference in the degree of membership of the feature vector of each collection point at the current moment in each field in different clusters; the correlation between any two types of data at the current moment and all historical moments of each collection point in each field is analyzed, and the reference reliability of the feature vector of the current moment of any collection point and the remaining collection points in each field is determined based on the difference in the correlation between any collection point and the remaining collection points in each field, and the difference between the same type of data of any collection point and the remaining collection points; Divide the various types of data at the current moment and all historical moments of each collection point in each field into various data clusters, and determine the abnormality degree of any type of data at the current moment of each collection point in each field based on the difference between the data cluster where any type of data is located at the current moment and the remaining data clusters of the any type of data; determine the first abnormal value of the feature vector of each collection point in each field at the current moment based on the abnormality degree of all types of data at the current moment of each collection point in each field; Based on all the reference reliabilities of any collection point in each field and the differences between the same type of data, determine the second abnormal value of the feature vector of each collection point in each field at the current moment; combine the first abnormal value and the second abnormal value to correct the initial abnormal value to obtain the corrected abnormal value of the feature vector of each collection point in each field at the current moment, obtain the processing results of all types of data of each collection point in each field at the current moment, and perform data fusion and digital processing; The processing result of obtaining all types of data at each collection point in each field at the current moment includes: Calculate the mean of the first outlier value and the second outlier value, the corrected outlier value is the product of the initial outlier value and the mean, if the corrected outlier value is greater than or equal to the preset outlier threshold, then remove all types of data in the feature vector corresponding to the current moment, otherwise, retain all types of data in the feature vector corresponding to the current moment.

2. The urban area model digitization method based on multi-source heterogeneous information collation as claimed in claim 1 is characterized in that: The process of determining the initial outlier value is as follows: For any feature vector of each collection point at the current moment, the membership of any feature vector in each cluster cluster is counted, the absolute value of the difference between the membership of any feature vector in the cluster cluster where it is located and the membership of the remaining cluster clusters is calculated, and the normalized value of the sum of all the absolute values ​​of the difference of any feature vector is used as the initial outlier value of any feature vector.

3. The urban area model digitization method based on multi-source heterogeneous information collation as claimed in claim 1 is characterized in that: The determination of the reference reliability includes: The current moment of each collection point in each field and all types of data at all historical moments are combined into each data sequence at the current moment, and the correlation between any two data sequences at the current moment of each collection point in each field is calculated, recorded as the first correlation, and the differences between all the first correlations of any collection point and the remaining collection points are analyzed to determine the parameter distribution similarity of the feature vectors at the current moment of any collection point in each field and the remaining collection points; Calculate the correlation between the data sequence of the same type at the current moment of any collection point in each field and the remaining collection points, record it as the second correlation, and take the normalized value of the sum of all the second correlations at the current moment of any collection point in each field and the remaining collection points as the parameter value similarity of the feature vector at the current moment of any collection point in each field and the remaining collection points; The product of the parameter distribution similarity and the parameter value similarity is used as the reference reliability.

4. The method for digitizing an urban area model based on multi-source heterogeneous information collation as claimed in claim 3, characterized in that: The parameter distribution similarity is calculated as follows: ; For collection point With collection point The similarity of the parameter distribution of the feature vector A at the current moment, For collection point With collection point of Type data, the absolute value of the difference of the first correlation between the types of data, is the number of types contained in the eigenvector A, and exp() represents the exponential function with a natural constant as the base.

5. The urban area model digitization method based on multi-source heterogeneous information collation as claimed in claim 1 is characterized in that: The process of determining the abnormality degree is as follows: Taking multiple elements connected in time series in each data cluster as a sequence segment, if the data of any type at each acquisition point at the current moment in each field is located in the sequence segment, calculating the absolute value of the difference between the number of sequence segments in the data cluster where the data of any type is located and the number of sequence segments in the remaining data clusters, recording it as the first absolute value of the difference, calculating the sum of all the first absolute values ​​of the difference of the data of any type, recording it as the first sum, calculating the metric distance between the sequence segment where the data of any type is located and the remaining sequence segments, calculating the sum of all the metric distances of the data of any type, recording it as the second sum; The product of the first sum value and the second sum value is used as an abnormal factor of any type of data at the current moment; Determine the abnormal significance of any type of data at the current moment based on the abnormal factor; The normalized value of the product of the abnormal factor and the abnormal significance is used as the abnormal degree. If any type of data is not located in the sequence segment, the abnormal degree is set to a preset value.

6. The method for digitizing an urban area model based on multi-source heterogeneous information collation as claimed in claim 5, characterized in that: The calculation method of the abnormal significance is: ; is the abnormal significance of any type of data at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing the o-th data in the sequence segment where any type of data is located at the current moment, is the absolute value of the difference between the abnormal factors calculated before and after removing any type of data from the sequence segment where any type of data is located at the current moment, and D is the number of data contained in the sequence segment where any type of data is located at the current moment.

7. The urban area model digitization method based on multi-source heterogeneous information collation as claimed in claim 1 is characterized in that: The first abnormal value is a normalized value of the sum of the abnormal degrees of all types of data at each collection point in each field at the current moment.

8. The urban area model digitization method based on multi-source heterogeneous information collation as claimed in claim 1 is characterized in that: The second outlier value is calculated as follows: ; For collection point The second outlier value of the feature vector A at the current moment, For collection point With collection point The reference reliability of the feature vector A at the current moment, M is the number of acquisition points, For collection point With collection point The absolute value of the difference between the h type data at the current moment, is the number of types contained in the feature vector A, and norm() is the normalization function.

9. An urban area model digitization system based on multi-source heterogeneous information collation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images