CN118013825A

CN118013825A - Multi-level multi-type city space-time collaborative design method, system, terminal and medium

Info

Publication number: CN118013825A
Application number: CN202410110516.5A
Authority: CN
Inventors: 赵志刚; 陈业滨; 杨佳豪; 马丁; 贺彪; 朱维; 颜雪
Original assignee: Shenzhen University
Current assignee: Shenzhen University
Priority date: 2024-01-25
Filing date: 2024-01-25
Publication date: 2024-05-10

Abstract

The invention provides a multi-layer multi-type city space-time collaborative design method, a system, a terminal and a medium, which particularly relate to the technical field of data processing, and the scheme comprises the following steps: constructing a space collaborative design ontology model by using a multi-level multi-type city planning ontology; based on a space collaborative design ontology model, extracting and predicting text of space planning resource information by using a deep neural network, so as to construct a space collaborative design knowledge graph; and carrying out space-time correlation on the time planning elements and the space collaborative design knowledge graph to obtain a space-time information knowledge graph correlation model, and carrying out dynamic evaluation and adjustment on the space-time information knowledge graph correlation model by utilizing multi-level and multi-type design indexes to obtain a city space-time collaborative design result. According to the scheme, the information contained in urban planning data can be fully mined, the design contradiction between planning data of different levels and different types is eliminated, and the efficiency and accuracy of urban space-time design can be improved.

Description

Multi-level multi-type city space-time collaborative design method, system, terminal and medium

Technical Field

The invention relates to the technical field of data processing, in particular to a multi-level multi-type city space-time collaborative design method, a system, a terminal and a medium.

Background

Space planning is an important measure for improving the space treatment capability and realizing the national development strategy, and is an important basis for future development, construction and management of cities. The city planning system in China has huge variety and complex content, and planning treatment involves different levels and types. Because each level of planning has respective specific target requirements, the target is often nonuniform, partial projects are overlapped in a crossing way, meanwhile, the data standard and the policy target are different, and the management and implementation process of the planning projects is severely restricted. Therefore, how to efficiently provide comprehensive and reliable collaborative planning information for the system is a problem to be considered and solved by the planning departments.

The coordination of space planning is a complex work, mainly embodied in the hierarchical linking of the upper and lower levels and the transverse coordination of the same level, and the linking, planning coordination and programming of the current planning still have a plurality of problems. The traditional urban space planning generally follows a top-down mode, namely, based on a set overall development target, planners perform corresponding space layout in different hierarchical scale areas, and the problems of inconsistent data base, unshared planning results, difficult constraint coordination and the like among different planning hierarchies in the city are easily caused. Meanwhile, the unified structural expression and semantic association description are lacking among planning data of different levels and different types, the characteristics and the relation of various planning contents are not fully mined, and hidden knowledge in the urban space planning field is difficult to find and express, so that the 'information island' problem is serious, and the complex space planning problem is difficult to deal with. Therefore, the prior art has the defects that urban space planning information is not fully mined, and planning information of various levels cannot be cooperatively utilized to carry out effective space planning.

Disclosure of Invention

In view of the shortcomings of the prior art, the invention aims to provide a multi-level multi-type urban space-time collaborative design method, a system, a terminal and a medium, which aim to solve the problems that urban space planning information is not fully mined and planning information of various levels cannot be cooperatively utilized to carry out effective space planning in the prior art.

In order to achieve the above object, a first aspect of the present invention provides a multi-level multi-type urban space-time collaborative design method, comprising:

Acquiring a plurality of types of urban planning knowledge bodies in a plurality of levels, and constructing a space collaborative design ontology model by utilizing the urban planning knowledge bodies;

Based on the space collaborative design ontology model, extracting and predicting text of space planning resource information by using a deep neural network to obtain space planning elements;

representing the space planning elements in a triplet form, and constructing a space collaborative design knowledge graph;

Acquiring a time planning element, and performing space-time correlation on the time planning element and the space collaborative design knowledge graph to acquire a space-time information knowledge graph correlation model;

and acquiring multi-level and multi-type design indexes, and dynamically evaluating and adjusting the space-time information knowledge graph correlation model by utilizing all the design indexes to acquire a city space-time collaborative design result.

Optionally, the constructing the spatial collaborative design ontology model by using the urban planning ontology includes:

Determining a space planning entity and a space planning entity relationship based on the content and the characteristics of the city planning ontology;

And constructing a space collaborative design ontology model according to preset attribute constraint conditions and specifications of the planning field by utilizing a logic and hierarchical relationship formed by the space planning entity and the space planning entity relationship.

Optionally, the extracting and text predicting the space planning resource information by using the deep neural network based on the space collaborative design ontology model to obtain a space planning element includes:

labeling the space planning resource information in the urban planning knowledge body based on the space collaborative design ontology model to obtain labeled space planning resource information;

based on the noted space planning resource information, learning by using a BERT word embedding layer in a deep neural network to obtain word vectors, segment vectors and position vectors;

extracting the relation among the captured word vectors, the fragment vectors and the position vectors by utilizing BiLSTM layers in the deep neural network to obtain initial space planning elements;

And screening the initial space planning elements by using a conditional random field layer in the deep neural network to obtain space planning elements.

Optionally, the extracting the relationship between the captured word vector, the segment vector and the position vector by using BiLSTM layers in the deep neural network to obtain an initial space planning element includes:

performing context prediction on the captured word vector, the captured fragment vector and the captured position vector by utilizing BiLSTM layers in the deep neural network to obtain context information;

Based on the context information, splicing the context information, the word vector, the fragment vector and the position vector by using an attention mechanism to obtain a prediction sequence;

And extracting a space planning entity and a space planning entity relation from the prediction sequence to obtain an initial space planning element.

Optionally, the expressing the space planning element in a form of a triplet, and constructing a space collaborative design knowledge graph includes:

extracting a space planning entity and a space planning entity relation in the space planning elements, and constructing a triplet by utilizing the space planning entity, the space planning entity relation and the attribute of the space planning entity;

And constructing a spatial collaborative design knowledge graph by utilizing all the triples.

Optionally, the performing space-time correlation between the time planning element and the spatial collaborative design knowledge graph to obtain a space-time information knowledge graph correlation model includes:

obtaining planning requirements of different levels in the time planning elements, and generating space position data based on the planning requirements;

Generating semantic relation data based on the spatial collaborative design knowledge graph;

Acquiring time node information in the time planning elements, and generating time relation data based on the time node information;

and obtaining a space-time information knowledge graph association model by using the space position data, the semantic relation data and the time relation data.

Optionally, after the obtaining the urban space-time co-design result, path analysis between space planning entities in the space planning element is further performed according to the urban space-time co-design result, which specifically includes:

determining a path between at least two space planning entities based on the urban space-time co-design result;

Determining at least one of correlation between each space planning entity, correlation degree and attribute of each space planning entity according to the paths;

And updating the urban space-time co-design result based on the correlation, the correlation degree and/or the attribute of the space planning entity.

A second aspect of the present invention provides a multi-level multi-type urban spatiotemporal collaborative design system, the system comprising:

the system comprises a space collaborative design ontology model construction module, a space collaborative design ontology model generation module and a model generation module, wherein the space collaborative design ontology model construction module is used for acquiring a plurality of types of urban planning ontology in a plurality of levels and constructing a space collaborative design ontology model by utilizing the urban planning ontology;

The prediction module is used for extracting and predicting text of space planning resource information by utilizing a deep neural network based on the space collaborative design ontology model to obtain space planning elements;

the space collaborative design knowledge graph construction module is used for expressing the space planning elements in a form of triples and constructing a space collaborative design knowledge graph;

The time-space information knowledge graph correlation model construction module is used for acquiring time planning elements, and performing time-space correlation on the time planning elements and the space collaborative design knowledge graph to acquire a time-space information knowledge graph correlation model;

and the multi-level multi-type city space-time collaborative design module is used for obtaining multi-level multi-type design indexes, and dynamically evaluating and adjusting the space-time information knowledge graph correlation model by utilizing all the design indexes to obtain a city space-time collaborative design result.

The third aspect of the present invention provides an intelligent terminal, which includes a memory, a processor, and a multi-hierarchy multi-type city space-time co-design program stored on the memory and executable on the processor, wherein the multi-hierarchy multi-type city space-time co-design program, when executed by the processor, implements the steps of any one of the multi-hierarchy multi-type city space-time co-design methods described above.

A fourth aspect of the present invention provides a computer readable storage medium having stored thereon a multi-level multi-type urban spatiotemporal co-design program which when executed by a processor implements the steps of any one of the multi-level multi-type urban spatiotemporal co-design methods described above.

Compared with the prior art, the beneficial effects of this scheme are as follows:

The invention utilizes a multi-level and multi-type urban planning knowledge ontology to construct a space collaborative design ontology model, and based on the space collaborative design ontology model, utilizes a deep neural network to extract and predict space planning resource information to obtain space planning elements, and expresses the space planning elements in a triplet form to construct a space collaborative design knowledge graph; and performing space-time correlation on the time planning elements and the space collaborative design knowledge graph to obtain a space-time information knowledge graph correlation model, and dynamically evaluating and adjusting the space-time information knowledge graph correlation model by utilizing multi-level and multi-type design indexes to obtain a city space-time collaborative design result. According to the scheme, the time and space data of the urban planning are effectively linked, information island barriers among the urban planning data can be avoided, the design contradiction among planning data of different levels and different types can be eliminated, and the efficiency and accuracy of urban space-time design can be remarkably improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a multi-level multi-type city space-time collaborative design method of the present invention;

FIG. 2 is a flow chart of the space planning entity identification of the present invention;

FIG. 3 is a schematic representation of the input representation of the BERT layer of the present invention;

FIG. 4 is a schematic representation of a fragment encoding representation of the BERT layer of the present invention;

FIG. 5 is a diagram of the BiLSTM-Attention model of the present invention;

FIG. 6 is a schematic diagram of a spatiotemporal information knowledge graph correlation model structure of the present invention;

FIG. 7 is a schematic diagram of a multi-level multi-type urban space-time co-design system architecture module according to the present invention;

fig. 8 is a schematic structural diagram of an intelligent terminal according to the present invention.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

It should be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.

The following description of the embodiments of the present invention will be made more fully hereinafter with reference to the accompanying drawings, in which embodiments of the invention are shown, it being evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.

The invention provides the following concept of network taxi sharing matching, and firstly, the body model construction is designed in a space collaborative mode. And constructing a space collaborative design ontology model by analyzing the space planning related data. The data entity and relationship are then extracted. The BERT-BiLSTM-CRF and BiLSTM-Attention models are used to extract entities and relationships of unstructured data. Thirdly, constructing a spatial collaborative design knowledge graph. And expressing the extracted entities and relations in the form of triples, and constructing a spatial collaborative design knowledge graph. Fourth, the time planning elements are associated and fused with the knowledge graph. And carrying out space-time correlation on the collaborative design elements, the historical data, the constraint indexes and the like by utilizing multi-level and multi-type space units, and fusing the space planning layers with space-time attributes with the knowledge maps to form a space planning element knowledge map correlation model with space-time characteristics. Finally, the multi-level multi-type urban space collaborative design is realized. The multi-level multi-type city space design collaboration based on the knowledge graph comprises the following steps: the method comprises the steps of intelligent retrieval of planning elements, collaborative analysis of the planning elements, conflict detection of planning contents and multi-level index collaboration. The method effectively solves the information island barriers among urban planning data, ablates the collaborative contradiction among multiple subjects, and improves the efficiency and accuracy of urban space-time design.

The embodiment of the invention provides a multi-level multi-type city space-time collaborative design method which is deployed on electronic equipment such as intelligent wearing equipment, computers, servers and the like, wherein an application scene is city time and space planning, and aims at the situation that planning elements of multi-level multi-type city time and space layers exist. Specifically, as shown in fig. 1, the steps of the method in this embodiment include:

step S100: acquiring a plurality of types of urban planning knowledge bodies in a plurality of levels, and constructing a space collaborative design ontology model by utilizing the urban planning knowledge bodies;

Specifically, because each level of each plan in the city planning has respective specific target requirements, and meanwhile, data standards have differences, for example, space planning resource information of the city planning comprises a plurality of levels of forestry offices, civil government offices, natural resource offices, water conservancy offices and the like, and each level comprises a plurality of types of data, for example, the forestry offices comprise scenic region gardens, urban and rural greening, natural ecological forest regions and the like.

Based on the method, the system and the device, unified planning is carried out by collecting various planning data of each level, and the various planning data of each level are classified to obtain a plurality of types of urban planning knowledge ontologies, wherein the ontologies are descriptions of clear specifications of a certain entity concept system, the description and definition of knowledge data by a knowledge graph are called knowledge ontologies or knowledge system, the ontologies are important knowledge bases, and the ontologies of the knowledge graph comprise types of objects, types of attributes and types of relations. The ontology of space types in urban planning can be divided into concepts of planning names, planning types, planning levels, planning indexes, planning deadlines, planning ranges, responsibility subjects, planning bases and the like. And then, utilizing the objects, attributes and relations of all the urban planning ontology to carry out collaborative expression, and constructing a spatial collaborative design ontology model.

Step S200: based on the space collaborative design ontology model, extracting and predicting text of space planning resource information by using a deep neural network to obtain space planning elements;

Specifically, space planning resource information is obtained, preprocessing is performed on the space planning resource information to remove invalid information, delete repeated information, sequence labeling and the like, wherein the sequence labeling refers to labeling tasks for each element or part of elements in a sequence, the task for each element needing to be labeled as a label is called as an original label, and the task for all segments labeled as the same label is called as a joint label. In the embodiment, the multi-level and multi-type text type space planning resource information is subjected to joint labeling, and the labeled space planning resource information is obtained. And extracting the marked space planning resource information by using the deep neural network to obtain a space planning entity and space planning entity relation, and performing context prediction based on the space planning entity and space planning entity relation to obtain a space planning element. It should be stated that the acquired space planning resource information may be text type data or non-text type data, and if the data is non-text type data, the data is converted into text type.

Step S300: representing the space planning elements in a triplet form, and constructing a space collaborative design knowledge graph;

specifically, various types of objects, attributes and relation information in the space planning elements are expressed in the form of triples, local knowledge maps are built by using corresponding triples of all types of data in each level, and then the local knowledge maps corresponding to all levels are fused to generate a multi-level multi-type space collaborative design knowledge map. The construction method of constructing the single-layer local knowledge patterns and fusing the local knowledge patterns of all the single layers not only reduces construction difficulty, but also can accurately and clearly reflect the association relation among the local knowledge patterns.

It should be stated that, a certain type of space planning data in a certain hierarchy may further include a plurality of sub-types of space planning data, and at this time, a sub-local knowledge graph may also be constructed by using all the sub-types of space planning data in the type, and participate in the fusion process. Obviously, the relationships among the local knowledge maps included in the constructed multi-level multi-type spatial collaborative design knowledge maps can be of the same level or different levels.

Step S400: acquiring a time planning element, and performing space-time correlation on the time planning element and the space collaborative design knowledge graph to acquire a space-time information knowledge graph correlation model;

specifically, in an actual project, elements of city planning involve not only space planning elements but also planning elements on a time axis, which are called time planning elements. In order to improve the rationality and feasibility of urban planning, the embodiment performs space-time correlation on the time planning elements and the space collaborative design knowledge graph to obtain a space-time information knowledge graph correlation model. For example, detailed information of element changes in a planning event can be recorded according to a time sequence, dynamic evolution of the whole planning scene is further described, an evolution relation of space planning elements is established, namely, the evolution relation of the space planning elements is described by utilizing the time planning elements, then space-time correlation is carried out on the evolution relation of the space planning elements and the space relation between the space planning elements in a space collaborative design knowledge graph, and the method has important significance for backtracking historical planning states by planning personnel and realizing overall process management of planning items.

Step S500: and acquiring multi-level and multi-type design indexes, and dynamically evaluating and adjusting the space-time information knowledge graph correlation model by utilizing all the design indexes to acquire a city space-time collaborative design result.

Specifically, in the implementation process of space-time planning, a planner can associate planning units corresponding to each type in each level through a space-time information knowledge graph association model, quickly look up the decomposition conditions of space-time planning indexes of different types in different levels through upper-lower level relations, and can perform the summarization calculation of indexes of each type in each level according to the index values of the layer node attributes in the knowledge graph to obtain the space-time planning indexes of each level and even each type. Comparing the calculated space-time planning index with a corresponding design index, and if the calculated space-time planning index does not accord with the design standard, dynamically evaluating and adjusting a space-time information knowledge graph correlation model to obtain a city space-time collaborative design result; if the design standard is met, the urban space-time collaborative design result is directly used.

In the embodiment, firstly, a spatial collaborative design ontology model is built by using a multi-level and multi-type urban planning ontology, and based on the spatial collaborative design ontology model, spatial planning resource information is extracted and text predicted by using a deep neural network to obtain spatial planning elements, and the spatial planning elements are expressed in a triplet form to build a spatial collaborative design knowledge graph; and performing space-time correlation on the time planning elements and the space collaborative design knowledge graph to obtain a space-time information knowledge graph correlation model, and dynamically evaluating and adjusting the space-time information knowledge graph correlation model by utilizing multi-level and multi-type design indexes to obtain a city space-time collaborative design result. According to the scheme, the time and space data of the urban planning are effectively linked, information island barriers among the urban planning data can be avoided, the design contradiction among planning data of different levels and different types can be eliminated, and the efficiency and accuracy of urban space-time design can be remarkably improved.

In one embodiment, the constructing a spatial collaborative design ontology model using the urban planning ontology in step S100 includes:

Step S110: determining a space planning entity and a space planning entity relationship based on the content and the characteristics of the city planning ontology;

Specifically, firstly, researching the current planning situation in the planning field and making planning business demands, and defining the knowledge body in the planning field and constructing the knowledge body in the field; according to the content and the characteristics of the knowledge ontology in the urban planning field, a concept system formed by core concepts and all concepts in the planning field is determined, the concepts are converted into space planning entities, and the relationships among the concepts implied in the concept system are converted into space planning entity relationships.

For example, the urban planning ontology is divided into 8 concepts of planning names, planning types, planning levels, planning indexes, planning deadlines, planning scopes, responsibility subjects and planning basis. The planning types comprise various special planning, overall planning, detailed planning and the like, the planning level comprises market level planning, district level planning and the like, the planning indexes comprise constraint indexes and regulation indexes, the responsibility main body comprises forestry offices, civil government offices, natural resource offices, water conservancy offices and the like, and the planning is based on technical standards and laws and regulations. Relationship attributes between partial concepts as shown in the spatial planning entity relationship definition table shown in table 1:

Table 1:

Step S120: and constructing a space collaborative design ontology model according to preset attribute constraint conditions and specifications of the planning field by utilizing a logic and hierarchical relationship formed by the space planning entity and the space planning entity relationship.

Specifically, according to preset attribute constraint conditions and specifications of the planning field, setting attributes and attribute values meeting planning design requirements for each space planning entity, and summarizing and arranging space planning entities and space planning entity relations to form logic and hierarchical relations among the space planning entities, so that a space collaborative design ontology model is built based on the logic and hierarchical relations.

In this embodiment, according to the content and characteristics of the ontology in the urban planning domain, the space planning entity and the space planning entity relationship in the planning domain are determined, and according to the attribute constraint condition of the space planning entity and the specification of the planning domain, the logic and hierarchical relationship formed between the space planning entity relationships are determined, so that a space collaborative design ontology model is constructed, and the consistency and timeliness of the ontology in the urban planning domain and the knowledge in the urban planning domain can be ensured.

In one embodiment, based on the spatial co-design ontology model in step S200, the extracting and text predicting the spatial planning resource information by using the deep neural network to obtain the spatial planning element includes:

step S210: labeling the space planning resource information in the urban planning knowledge body based on the space collaborative design ontology model to obtain labeled space planning resource information;

Specifically, BIO is commonly used as a kind of joint labeling method to solve the problem of information extraction in the natural language processing, and specifically B, I, O represents Begin Inner Other respectively. Further, B-X indicates that the entity is of the X type and is located at the beginning of the entity fragment, I-X indicates that the entity is of the X type and is located in the middle of the entity fragment, and O indicates that the element is not of the X type.

In the embodiment, a BIO labeling method is adopted to carry out entity labeling on multi-level and multi-type planning text data. B-X, I-X represents specific entity categories, wherein X is NAME, DATE, RAN, LEV, TYPE, IND, BAS, RES and is respectively represented as a planning name, a planning time, a planning range, a planning level, a planning type, a planning index, a planning basis and a responsibility main body.

For example, a detailed planning of a sea area in a certain city, the scope of which includes a coastal zone region composed of the sea area and an adjacent land area, is mainly based on the "use management regulations for the sea area in the certain city", the planning is responsible for explanation by a marine administrative authority, and table 2 shows an example text marked by using a BIO marking method.

Table 2:

relationships generally included in the planning text mainly include "reconciliation", "collusion", "inclusion", "correlation", "scope", "planning basis", and "responsibility main", etc. Table 3 shows a specific definition example of the relationship of the planning text.

Table 3:

Step S220: based on the noted space planning resource information, learning by using a BERT word embedding layer in a deep neural network to obtain word vectors, segment vectors and position vectors;

step S230: extracting the relation among the captured word vectors, the fragment vectors and the position vectors by utilizing BiLSTM layers in the deep neural network to obtain initial space planning elements;

step S240: and screening the initial space planning elements by using a conditional random field layer in the deep neural network to obtain space planning elements.

Specifically, the deep neural model adopted for entity recognition of the annotated space planning resource information in the embodiment is an integral structure formed by combining a BERT word embedding layer, a BiLSTM layer and a conditional random field model (CRF) layer, and a planning knowledge entity recognition flow chart is shown in fig. 2.

The main function of the BERT word embedding layer is to convert each word into a computer recognizable vector form, and this vector representation is passed as a parameter to the subsequent BiLSTM and CRF layers. The BERT model is constructed on the basis of a bidirectional transducer model, so that more accurate and comprehensive language representation can be learned, and the problems of word ambiguity and the like can be solved. The input representation of BERT consists of the sum of word vectors, segment vectors and position vectors, as shown in fig. 3, so that the BERT word embedding layer can better combine context information in text and the relationship between words and sentences, thereby extracting more feature information. The word vector is the vector representation of the current word, the word vector of the BERT is learned in the training process, each word is represented by a vector with a fixed size, the word vector comprises two flag bits, namely [ CLS ] and [ SEP ], wherein [ CLS ] is placed at the first position of a sentence and used for the subsequent classification task, and [ SEP ] is used for dividing two sentences. The segment vector is used for encoding a sentence to which the current word belongs, and the BERT layer can randomly combine two different sentences as a model input in the training process, so that the model can learn, understand and generate context information, and the reasoning capacity of the model is enhanced. For example, for the input of "a certain region is located in Shenzhen city" and "a certain region formulates a public service facility plan" sentence pair, fragment encoding is as shown in FIG. 4. The position vector is a fixed vector representing the absolute position of each word in the sentence. Text data often has timing characteristics, i.e., the positions of words in text are different, and the meaning of the expressions may be different, so that position information in a sequence is better captured by adjusting position vectors in the BERT layer training process.

After the BERT layer inputs the vector consisting of the sum of the word vector, the segment vector and the position vector, different downstream tasks can be well adapted through a pre-training process. The main tasks of the pretraining process of BERT include: mask language model pre-training and next sentence prediction tasks.

The Mask language model pre-training task is to Mask (Mask) words with preset proportion (such as 15%) in the input text, and then predict the masked original values through the information provided in the context. For example, for "a specific public facility plan is formulated in a certain area", masking is performed on the term "specific" in the specific item to obtain a masked sentence: "a certain zone makes a utility mask plan". In order to avoid the problem that random masking can cause that certain words cannot appear in a downstream fine tuning stage, so that inconsistency between pre-training and fine tuning is caused, in the downstream fine tuning stage, masking is performed by reducing masking probability of a preset proportion, or the masked words are replaced by other words by a preset proportion, or the original words are displayed by a preset proportion. For example, for a masked sentence: "a certain area makes a public facility [ mask ] [ mask ] plan", measures taken in the downstream fine tuning stage are: and (3) performing normal masking processing according to the probability of 80%, namely making a public facility [ mask ] [ mask ] plan for a certain area by a sentence to be processed in the fine tuning stage, replacing the masked word with an arbitrary word according to the probability of 10%, namely making a public facility green land plan for the certain area by the sentence to be processed in the fine tuning stage, and performing no masking processing on the word to be masked according to the probability of 10%, namely making a public facility private land plan for the certain area by the sentence to be processed in the fine tuning stage.

The next sentence prediction task is to predict the next sentence of the text information. Since the relation between two pieces of text cannot be learned by only relying on the masking task, the relation between two pieces of text is constructed by introducing the next sentence prediction task at the BERT layer. For example, two sentences are randomly selected, sentence a: a certain zone makes a utility mask plan. Sentence B: the [ mask ] [ mask ] planning period is 2020 to 2035, and whether the sentence A is the next sentence of the sentence B is predicted by a trained model.

The Bi-directional Long Shot Term Memory, biLSTM neural network can use forward LSTM and backward LSTM to capture forward information and backward information in the sequence, and splice the captured information together and access the output layer (i.e., C1, C2...c5) for prediction. Therefore, in the urban planning knowledge extraction, the embodiment utilizes BiLSTM layers in the deep neural network to extract the relation among the captured word vectors, the segment vectors and the position vectors so as to output the initial space planning elements, so that the characteristics of the context can be automatically captured, and the text prediction capability can be enhanced.

Conditional Random Fields (CRFs) are a common class of sequence labeling models that combine the advantages of the maximum entropy model and the hidden markov model, and are commonly used in named entity recognition tasks. Compared with other sequence labeling models, the CRF has the advantage of being capable of considering global information, namely modeling the whole sequence as a whole, so that the context characteristic information can be captured. Therefore, in this embodiment, the CRF layer is used as the downstream model of BiLSTM to screen the initial space planning element according to the constraint condition, so as to obtain the space planning element, which can effectively solve the problem of error that is not in line with logic between the output of each marking result in the sequence marking task. For example, constraints for setting the CRF layer are: the B-Name tag may be followed by I-Name, O, etc., but it is illegal to follow the I-RANGE or I-BAS, etc.

In one embodiment, the extracting the relationships between the captured word vector, the segment vector and the position vector by using the BiLSTM layers in the deep neural network in step S230 to obtain an initial space planning element includes:

Step S231: performing context prediction on the captured word vector, the captured fragment vector and the captured position vector by utilizing BiLSTM layers in the deep neural network to obtain context information;

step S232: based on the context information, splicing the context information, the word vector, the fragment vector and the position vector by using an attention mechanism to obtain a prediction sequence;

step S233: and extracting a space planning entity and a space planning entity relation from the prediction sequence to obtain an initial space planning element.

Specifically, for the complicated unstructured space planning text data, the embodiment adopts BiLSTM and Attention mechanism combined method to extract and integrate the space planning entity relationship, and the BiLSTM-Attention model structure is shown in fig. 5. The BiLSTM-attribute model based on the deep neural network can learn the bottom features by carrying out information mining from large-scale text data, so that the relation among entities is automatically extracted, and the efficiency and objectivity of extracting the relation of the space planning entities can be improved. Because the BERT embedding layer can ensure the correct ordering of words in the text and obtain the sentence-level characterization capability, and has strong language characterization capability and feature extraction capability, the embodiment adopts the BERT model based on the deep neural network to generate word vectors to be embedded into the BiLSTM layer, then carries out context prediction through bidirectional LSTM at the BiLSTM layer so as to capture the context information of the words, and simultaneously splices the context information, the word vectors, the fragment vectors and the position vectors through the attention mechanism so as to obtain a prediction sequence, so as to adaptively determine the keywords in the sentences of the text, and reasonably allocate weights so as to obtain effective information and reduce the influence of noise data. Finally, a classifier (such as a Softmax function) is used to extract the spatial planning entity relationship from the predicted sequence, so as to obtain an initial spatial planning element and output the initial spatial planning element as a result.

It should be noted that, as can be seen from comparing fig. 2 and fig. 5, the relation extraction model of the space planning text data is similar to the named entity recognition model, and is different in that the BiLSTM layers of the relation extraction model are connected with attention mechanisms, that is, the weight matrix of each word in the sentence is obtained by adding the attention model of the word level, so that high-value information can be rapidly screened and emphasized.

In one embodiment, the step S300 of representing the space planning element in the form of a triplet, and constructing a space collaborative design knowledge graph includes:

Step S310: extracting a space planning entity and a space planning entity relation in the space planning elements, and constructing a triplet by utilizing the space planning entity, the space planning entity relation and the attribute of the space planning entity;

step S320: and constructing a spatial collaborative design knowledge graph by utilizing all the triples.

Specifically, the space planning entity and space planning entity relation in the space planning elements are extracted, the entity and relation data are sequentially organized into a triplet form of a resource description framework (Resource Description Framework, RDF), such as [ entity, relation, entity ], or [ entity, attribute value ], a space collaborative design knowledge graph is constructed, then the space collaborative design knowledge graph is stored in a graph database (such as Neo 4J) through a storage engine, storage of the knowledge graph data is achieved, storage efficiency of tree structure data can be improved, and query, update, creation and deletion operations of the stored data are facilitated to be rapidly achieved.

In one embodiment, in step S400, performing space-time correlation on the time planning element and the spatial co-design knowledge graph to obtain a space-time information knowledge graph correlation model, which includes:

Step S410: obtaining planning requirements of different levels in the time planning elements, and generating space position data based on the planning requirements;

step S420: generating semantic relation data based on the spatial collaborative design knowledge graph;

step S430: acquiring time node information in the time planning elements, and generating time relation data based on the time node information;

Step S440: and obtaining a space-time information knowledge graph association model by using the space position data, the semantic relation data and the time relation data.

Specifically, the structure of the spatiotemporal information knowledge graph correlation model in this embodiment is shown in fig. 6, and includes: spatial location data, planning element data, and time nodes. The association relation involved in constructing the model is of three types: the first category is spatial position data, mainly the association between multi-level planning units, which is commonly used for analyzing and viewing between different levels; the second category is planning element data, which mainly comprises the association between each planning entity and the attribute and data thereof, and the association relationship between different planning entities; the third category is the association between planning elements, spatial locations and time nodes, which is commonly used to analyze the planning elements and interrelationships of spatial units involved at different times.

An exemplary example of constructing a spatiotemporal knowledge-graph correlation model is as follows:

And acquiring a time planning element, wherein the time planning element at least comprises planning units of different levels and time node information planned by each planning unit. Constructing association relations among planning units of different levels by using the planning units of each level to generate spatial position data; acquiring a space planning entity in a space collaborative design knowledge graph and attributes and data in each space planning entity, establishing an association relationship between the attributes and the data in each space planning entity and an association relationship between different space planning entities, and generating semantic relationship data; then establishing an association relationship among the space position data, the semantic relationship data and the time node information to generate time relationship data; and finally, obtaining a space-time information knowledge graph association model by using the spatial position data, the semantic relation data and the time relation data.

In this embodiment, detailed information of element changes in a planning event is recorded according to a time sequence, so that dynamic evolution of the whole planning scene is described, an evolution relationship of planning elements is established, and the method has important significance for planning personnel to trace back past planning states and realize overall process management of planning items.

In one embodiment, in step S500, a multi-level and multi-type design index is obtained, and all the design indexes are used to dynamically evaluate and adjust the spatio-temporal information knowledge graph association model to obtain a result of urban spatio-temporal collaborative design, which specifically includes:

When planning and compiling on a space plot, a designer often needs to refer to relevant planning data of the plot position, comprehensively understand and master the conditions of planning types, planning indexes, planning texts and the like related on the plot, and the association constraint relation of the planning types, so that benefits in all aspects are weighed, reasonable space layout is realized, and meanwhile, space distribution conditions of all planning types are also required to be checked, so that differentiated transformation strategies are provided for different development stages, existing problems and transformation requirements. Therefore, the invention can inquire the hierarchical land and feed back the planning knowledge graph on the land to the user, thereby meeting the planning business cooperation requirement corresponding to the multi-hierarchical and multi-type design indexes.

In one embodiment, after obtaining the urban space-time co-design result in step S500, step S600 is further included, where path analysis between space planning entities in the space planning element is performed according to the urban space-time co-design result, and specifically includes:

step S610: determining a path between at least two space planning entities based on the urban space-time co-design result;

step S620: determining at least one of correlation between each space planning entity, correlation degree and attribute of each space planning entity according to the paths;

Step S630: and updating the urban space-time co-design result based on the correlation, the correlation degree and/or the attribute of the space planning entity.

Specifically, based on the knowledge graph corresponding to the urban space-time collaborative design result, a starting node and a target node are selected, and a path between two space planning entities is determined according to the starting node and the target node to determine the correlation and the correlation degree between the two space planning entities, and a hidden relation between the two space planning entities or the attribute of the unknown entity can be further mined, for example, whether a certain space planning entity is a subclass or a member of another space planning entity is determined according to the mined hidden relation, so that the semantic relation between the space planning entities can be further understood. Through path analysis, the loss and the error in the knowledge graph can be found, and the quality of the knowledge graph can be further improved.

Further, after the urban space-time collaborative design result is obtained, historical planning data management, intelligent searching of planning elements, planning content space conflict detection, multi-level planning index conduction analysis and the like can be performed, and the method specifically comprises the following steps:

1) History planning data management:

Most of the urban planning information presents vivid time-varying characteristics, and planning staff need to conduct detailed research based on all current state information and various historical planning information when planning a planning scheme. According to the embodiment, the planning data are organized in series according to the time line to integrate the associated historical data, so that the whole process management of linkage of the current state of planning approval of the current generation of planning project data is realized.

2) Intelligent searching of planning elements:

And the intelligent searching of the planning elements is to search the planning entities and the relations thereof in the graph database and present the searched results to the user in the form of a graph structure. Based on the knowledge graph constructed by the invention, the association relations among different planning types, elements, texts and indexes can be queried, and the association relations are displayed in the form of a graph, for example, when the special planning of public service facilities is carried out, the upper planning, related planning, conductive management indexes and planning standard basis of the planning and the relations among the planning can be queried in a system, and the level and the type of the planning to be coordinated are positioned, so that planning staff can fully consider various information related to the planning staff during planning and compiling, and the cooperative efficiency of the planning staff is further improved.

3) Planning content space conflict detection:

In the process of realizing coordination of multiple types of planning elements, space conflict detection analysis is a crucial step, on one hand, the current planning, upper layer planning, space management and control lines and the like can be compared and analyzed to check whether the planning meets the requirements, and meanwhile, whether various planning space pattern spots on a space plot overlap or conflict can be detected, so that a planner is helped to coordinate and integrate different types of planning elements better, and more reasonable planning judgment is made.

Taking the city update unit planning as an example, the city update unit planning is to satisfy the following basic principles of planning constraints: (1) The method is not in conflict with various space control plans, namely the control line range, such as an ecological environment control line, a basic farmland protection area, a drinking water protection area and the like, cannot be broken through; (2) The upper planning is required to be obeyed, for example, the updating planning of a certain area is required to be obeyed to the overall planning of province and city to which the area belongs; (3) The special plans on the updating unit need to be coordinated with each other, so that conflicts between different plans are avoided, the space contradiction of the various plans is helped to be coordinated, and reasonable allocation of space resources is realized.

4) Conducting analysis of multi-level planning indexes:

In the implementation process of the homeland space planning, the planning indexes are conducted, implemented and executed layer by layer, so that the planning method is beneficial to promoting overall coordination in space among different layers of planning. For example, by checking the index decomposition conditions of different levels, or performing index summarization calculation of each level according to index values of layer node attributes in the knowledge graph, the real-time monitoring of the control indexes of different levels is realized.

According to the embodiment, the urban space-time collaborative design results are analyzed by setting different dimensions, so that the rationality and the feasibility of the urban space-time collaborative design results can be grasped dynamically, and meanwhile, the change condition of each dimension of the planned urban space-time collaborative design results can be tracked dynamically.

As shown in fig. 7, corresponding to the above-mentioned multi-level multi-type city space-time co-design method, the embodiment of the present invention further provides a multi-level multi-type city space-time co-design system, where the multi-level multi-type city space-time co-design system includes:

the space collaborative design ontology model construction module 710 is configured to obtain a plurality of types of urban planning ontologies in a plurality of levels, and construct a space collaborative design ontology model by using the urban planning ontologies;

the prediction module 720 is configured to extract and text predict the space planning resource information by using the deep neural network based on the spatial collaborative design ontology model, so as to obtain a space planning element;

The spatial collaborative design knowledge graph construction module 730 is configured to represent the spatial planning elements in a triplet form, and construct a spatial collaborative design knowledge graph;

The spatiotemporal information knowledge graph correlation model construction module 740 is used for acquiring time planning elements, and carrying out spatiotemporal correlation on the time planning elements and the spatial collaborative design knowledge graph to acquire a spatiotemporal information knowledge graph correlation model;

The multi-level multi-type city space-time collaborative design module 750 is configured to obtain multi-level multi-type design indexes, dynamically evaluate and adjust the space-time information knowledge graph correlation model by using all the design indexes, and obtain a city space-time collaborative design result.

Specifically, in this embodiment, the specific function of the above-mentioned multi-level multi-type city space-time collaborative design system may refer to the corresponding description in the above-mentioned multi-level multi-type city space-time collaborative design method, which is not described herein again.

Based on the above embodiment, the present invention further provides an intelligent terminal, and a functional block diagram thereof may be shown in fig. 8. The intelligent terminal comprises a processor, a memory, a network interface and a display screen which are connected through a system bus. The processor of the intelligent terminal is used for providing computing and control capabilities. The memory of the intelligent terminal comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a multi-level multi-type city spatiotemporal co-design program. The internal memory provides an environment for an operating system in a non-volatile storage medium and for the operation of a multi-level multi-type city space-time co-design program. The network interface of the intelligent terminal is used for communicating with an external terminal through network connection. The multi-level multi-type city space-time collaborative design program, when executed by the processor, implements the steps of any one of the multi-level multi-type city space-time collaborative design methods described above. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen.

It will be appreciated by those skilled in the art that the schematic block diagram shown in fig. 8 is merely a block diagram of a portion of the structure associated with the present inventive arrangements and is not limiting of the smart terminal to which the present inventive arrangements are applied, and that a particular smart terminal may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.

In one embodiment, an intelligent terminal is provided, where the intelligent terminal includes a memory, a processor, and a multi-level multi-type city space-time co-design program stored on the memory and capable of running on the processor, where the multi-level multi-type city space-time co-design program implements the steps of any one of the multi-level multi-type city space-time co-design methods provided by the embodiments of the present invention when executed by the processor.

The embodiment of the invention also provides a computer readable storage medium, wherein the computer readable storage medium is stored with a multi-level multi-type city space-time collaborative design program, and the steps of any multi-level multi-type city space-time collaborative design method provided by the embodiment of the invention are realized when the multi-level multi-type city space-time collaborative design program is executed by a processor.

It should be understood that the sequence number of each step in the above embodiment does not mean the sequence of execution, and the execution sequence of each process should be determined by its function and internal logic, and should not be construed as limiting the implementation process of the embodiment of the present invention.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present invention. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other manners. For example, the apparatus/terminal device embodiments described above are merely illustrative, e.g., the division of the modules or units described above is merely a logical function division, and may be implemented in other manners, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art will understand that; the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions are not intended to depart from the spirit and scope of the various embodiments of the invention, which are also within the spirit and scope of the invention.

Claims

1. The multi-level multi-type city space-time collaborative design method is characterized by comprising the following steps of:

2. The multi-level multi-type city space-time co-design method of claim 1, wherein said constructing a space co-design ontology model using said city planning ontology comprises:

3. The multi-level multi-type city space-time co-design method according to claim 1, wherein the extracting and text predicting the space planning resource information by using the deep neural network based on the space co-design ontology model to obtain the space planning element comprises:

4. The multi-level multi-type city space-time co-design method of claim 3, wherein said extracting relationships of said captured word vector, said segment vector, and said location vector using BiLSTM layers in said deep neural network to obtain an initial space-planning element comprises:

5. The multi-level multi-type city space-time co-design method according to claim 1, wherein the representing the space planning elements in the form of triples, constructing a space co-design knowledge graph, comprises:

6. The multi-level multi-type city space-time co-design method according to claim 1, wherein the performing space-time correlation between the time planning element and the space co-design knowledge graph to obtain a space-time information knowledge graph correlation model comprises:

7. The multi-level multi-type city space-time co-design method according to claim 1, further comprising, after the obtaining the city space-time co-design result, performing a path analysis between space planning entities in the space planning element according to the city space-time co-design result, specifically comprising:

8. A multi-level multi-type urban space-time co-design system, the system comprising:

9. A smart terminal comprising a memory, a processor, and a multi-level multi-type city spatiotemporal co-design program stored on the memory and operable on the processor, the multi-level multi-type city spatiotemporal co-design program when executed by the processor implementing the steps of the multi-level multi-type city spatiotemporal co-design method of any of claims 1-7.

10. A computer readable storage medium, wherein a multi-level multi-type city spatiotemporal co-design program is stored on the computer readable storage medium, which when executed by a processor, implements the steps of the multi-level multi-type city spatiotemporal co-design method of any of claims 1-7.