Building investigation object generation method, building investigation method, facility and system

Abstract

The present invention provides a building investigation object generation method, a building investigation method, a building investigation facility, and a system. The building investigation object generation method includes generating, for at least one investigation waiting system in a building, a plurality of investigation objects corresponding to the at least one investigation waiting system by a duplicate elimination scheme. The plurality of generated investigation objects are different from each other, and the duplicate elimination scheme is implemented at least on the basis of a correspondence relation between the investigation waiting system and the investigation object. It is thereby made possible to improve the accuracy and versatility of identifying the object to be investigated, improve the investigation efficiency, and save the investigation cost.

Description

Building survey target generation method, building survey method, equipment and system

[0001] The present invention relates to the field of building inspection, and in particular to a building inspection object generation method, a building inspection method, a building inspection facility and a system.

[0002] Currently, the market demand for intelligent buildings is gradually increasing, and the number of projects is also increasing accordingly. For building contractors, in order to be able to handle more projects, it is extremely important to complete the survey phase of the project efficiently and at low cost. In particular, the renovation of existing buildings to make them intelligent is time-pressured, so it is even more important to improve efficiency.

[0003] Before estimating or implementing the intelligentization of a building, it is usually used to investigate the facilities and environment of the project site and to predict the feasibility and amount of work required for later implementation.

[0004] For building systems, one conventional investigation method is to automatically push investigation tasks to corresponding investigators according to user needs, and the investigators investigate the investigation objects based on the investigation tasks. Another conventional investigation method is to first identify target investigation objects and then push the target investigation objects to investigators, so that the investigators only need to investigate the target investigation objects. For example, a method for identifying target investigation objects is for an expert to convert user needs into target investigation objects. In the field of communication investigation, for example, a predetermined task template library and user needs are used to identify corresponding target investigation objects.

[0005] Furthermore, when it is decided to inspect a target, the target is usually found and inspected based on a reference image of the target. In a conventional method for obtaining a reference image of the target, an engineer directly provides the reference image of the target. In another conventional method for obtaining a reference image of the target, a professional engineer manually selects a reference image that matches the site from a library of historical images of the target.

[0006] It should be noted that the above description of the technical background is only provided to make the technical solutions of the present invention clear, easy to explain completely, and easy to understand for those skilled in the art, and these solutions are only described in the background of the present invention, and are not considered to be known to those skilled in the art.

[0007] The inventor discovered that with the conventional method of directly pushing investigation tasks to investigators, the corresponding investigation targets are not pushed at the same time, so the investigator may have to identify the investigation targets that need to be investigated himself, and if the investigator is not familiar with the investigation targets, there is a risk that the investigation targets will be overlooked, resulting in low investigation efficiency; alternatively, if the investigator investigates all investigation targets directly, overlooking investigation targets can be avoided, but there are many investigation targets within a building, which requires considerable manpower and time, resulting in low investigation efficiency.

[0008] The conventional method of first identifying target survey targets and then pushing them to the surveyor significantly reduces the workload of the surveyor because the target survey targets are a narrowed-down subset of all survey targets within the building, but the conventional method of identifying target survey targets also has some flaws.

[0009] For example, the traditional method of expert-based translation of user needs into target research objects requires the expert's services, resulting in high labor and travel costs, low manual translation efficiency, and reduced research efficiency. Furthermore, manually identifying research objects also leads to overlooking of research objects. The traditional method of identifying research objects using a pre-defined task template library and user needs (in the field of telecommunications research) uses a task template to identify research objects based on individual and simple research objectives. This method is not suitable for complex, related research objectives (e.g., building systems). Furthermore, different user needs correspond to different research objects, making it difficult to pre-create task templates for each situation. This means that the task templates are not very versatile, and the accuracy of the identified research objects is also low. Furthermore, when multiple research tasks are performed simultaneously and the multiple research tasks have related research objects, the research objects identified by the traditional method will be duplicated and redundant. Furthermore, due to the lack of a clear way to express the relationship between each research object, researchers may need to repeatedly travel to the same location to investigate different research objects within the location, resulting in low research efficiency.

[0010] After identifying the survey object, if an engineer surveys the survey object based on a reference image of the survey object, the conventional method in which the engineer directly provides the reference image of the survey object may result in different aspects of the survey object due to the diversity of buildings, and the reference image of the survey object provided by the engineer may not match the survey site, making it difficult for the investigator to quickly find the survey object and resulting in low survey efficiency. The conventional method in which reference images suitable for the survey site are manually selected from a historical image library may result in limited images collected in the historical image library due to the diversity of buildings, making it difficult to include images of various survey sites, and the reference image obtained in this manner may not match the survey site and resulting in low survey efficiency. On the other hand, the method of manually selecting reference images relies on professional engineers, which results in high labor costs and low efficiency.

[0011] In order to solve one or more of the above problems, embodiments of the present invention provide a building survey object generation method, a building survey method, a building survey equipment, and a system that can improve the accuracy and versatility of identifying survey objects, solve the problem of overlapping and redundancy of survey objects, improve survey efficiency, and reduce survey costs.

[0012] According to a first aspect of an embodiment of the present invention, there is provided a building survey object generation method, the method including, for at least one system in a building to be surveyed, generating a plurality of survey objects corresponding to the at least one system to be surveyed using a deduplication method, the plurality of survey objects generated being different from each other, and the deduplication method being realized based at least on a correspondence relationship between the systems to be surveyed and the survey objects.

[0013] According to a second aspect of the present invention, there is provided a building survey method, the method comprising: generating a survey object using any of the building survey object generation methods described in the first aspect of the present invention; and surveying the generated survey object.

[0014] According to a third aspect of an embodiment of the present invention, there is provided a building inspection facility, the building inspection facility including a memory having a computer program stored therein, and a processor that, when executing the computer program, causes any one of the building inspection target generation methods described in the first aspect of an embodiment of the present invention to be realized, and / or any one of the building inspection methods described in the second aspect of an embodiment of the present invention to be realized.

[0015] According to a fourth aspect of the present invention, there is provided a building inspection system, the building inspection system including the building inspection equipment according to the third aspect of the present invention.

[0016] One of the beneficial effects of the embodiment of the present invention is as follows.

[0017] By automatically generating multiple different survey targets using a duplicate elimination method for systems in a building awaiting investigation, the problem of overlapping and redundant survey targets can be solved, which is advantageous for improving survey efficiency and reducing survey costs.In addition, by generating a survey target for at least one system awaiting investigation, it is necessary to conduct surveys on multiple systems awaiting investigation simultaneously, and in cases where multiple systems awaiting investigation have related or identical survey targets, it is possible to generate survey targets for the multiple systems awaiting investigation that do not overlap with each other.Furthermore, investigators can perform survey work on multiple systems awaiting investigation at once and do not need to repeatedly survey any of the survey targets, which is advantageous for further improving survey efficiency.

[0018] Furthermore, in the building survey object generation method according to the first aspect of the embodiment of the present invention, the step of "generating a plurality of survey objects corresponding to the at least one system to be surveyed using a deduplication method" includes outputting a plurality of survey objects corresponding to the at least one system to be surveyed using a pre-trained survey object generation / deduplication model, and the survey object generation / deduplication model is trained based on at least the correspondence between the systems to be surveyed and the survey objects and pre-set survey object deduplication rules.

[0019] This allows for the efficiency and accuracy of generating survey targets to be improved by identifying the survey targets using the survey target generation and deduplication model, thereby improving the efficiency of building surveys.

[0020] Furthermore, the survey target generation / duplication elimination model is trained based on the correspondence between systems awaiting investigation and survey targets, predetermined survey target duplication elimination rules, and predetermined association relationships between the survey targets, and generating multiple survey targets corresponding to at least one system awaiting investigation using the duplication elimination method includes outputting the multiple survey targets and the association relationships between the multiple survey targets using the survey target generation / duplication elimination model.

[0021] As a result, the survey target generation / duplication elimination model not only outputs survey targets, but also outputs the relationships between each survey target, which helps researchers to focus their research on multiple related survey targets, such as different survey targets in the same location, and avoids researchers having to repeatedly travel to the same location, which is advantageous for improving survey efficiency.

[0022] Furthermore, the search target generation and deduplication model includes a graph neural network model.

[0023] The graph neural network model can not only learn the correspondence between systems waiting to be investigated and investigation targets, but also the association relationships between investigation targets. Therefore, by using the investigation target generation and duplication elimination model of the present application, it is possible to identify investigation targets and the association relationships between investigation targets, thereby guiding the investigator's investigation work and improving investigation efficiency.

[0024] Furthermore, the building survey object generation method further includes training a survey object generation and deduplication model, and training the survey object generation and deduplication model includes training an initial model using previously acquired training samples to obtain the survey object generation and deduplication model, and the training samples include at least systems to be surveyed, survey objects, and correspondence between the systems to be surveyed and the survey objects.

[0025] This allows corresponding investigation targets to be output based on the systems awaiting investigation based on an investigation target generation and deduplication model trained from the systems awaiting investigation, investigation targets, and the correspondence between the systems awaiting investigation and investigation targets, thereby improving the efficiency and accuracy of investigation target generation.

[0026] Furthermore, generating a plurality of survey targets corresponding to the at least one system to be investigated using the deduplication method includes identifying survey targets corresponding to each system to be investigated in the at least one system to be investigated based on a pre-generated lookup table, and deduplicating all of the identified survey targets to obtain a plurality of survey targets corresponding to the at least one system to be investigated, wherein the lookup table includes at least a correspondence relationship between the system to be investigated and the survey targets.

[0027] This allows the use of a lookup table to identify the survey object, making it relatively easy to implement the lookup table method, and allowing accurate survey objects to be obtained quickly, thereby improving the efficiency and accuracy of survey object generation and improving the efficiency of building surveys.

[0028] Furthermore, the lookup table further includes association relationships between the survey objects, and the building survey object generation method further includes identifying association relationships between the plurality of survey objects based on the lookup table.

[0029] This allows the use of a lookup table to identify survey subjects and the relationships between each survey subject, which helps researchers concentrate on surveying multiple survey subjects that are related to each other, such as different survey subjects in the same location, and avoids researchers having to repeatedly travel to the same location, which is advantageous in improving survey efficiency.

[0030] Furthermore, the association relationship between the study objects includes at least one of an electrical relationship, a mechanical relationship, and a spatial positional relationship between the study objects.

[0031] This helps the researcher quickly understand the electrical, mechanical, and spatial relationships between each of the survey objects that require investigation, which is advantageous for understanding the relationships between survey objects early on and improving survey efficiency.

[0032] Furthermore, the building survey object generation method further includes determining a survey order for the plurality of survey objects based on association relationships between the survey objects.

[0033] This allows for a rational arrangement of the survey order, and provides researchers with a reference for the order in which to survey each survey subject, which is advantageous for improving survey efficiency.

[0034] Additionally, the building survey object generation method further includes representing the plurality of survey objects with at least one of a picture, a video, text, a map, and a three-dimensional model.

[0035] This provides various display modes of the survey object according to actual needs, and makes it easier for the surveyor to see the survey object, which is advantageous in improving survey efficiency.

[0036] Furthermore, the building survey target generation method further includes identifying the at least one system to be surveyed according to user needs, wherein identifying the at least one system to be surveyed according to user needs includes obtaining a user needs table including a plurality of systems to be surveyed and user markers for the systems to be surveyed, and identifying the systems to be surveyed marked by the user as the at least one system to be surveyed.

[0037] This allows systems to be identified based on the user's needs, and then identifies research targets based on the identified systems to be investigated, eliminating the need to manually convert the user's needs into systems to be investigated, which is advantageous in improving the efficiency of generating research targets and further improving research efficiency.

[0038] Furthermore, the building survey object generation method further includes formatting the needs table to obtain graph data corresponding to the at least one system to be surveyed.

[0039] This allows the user needs of the graph data format to be directly input into the search object generation and deduplication model trained based on the graph neural network to obtain the corresponding search objects, which is advantageous for improving the efficiency of search object generation and further improving the search efficiency.

[0040] Furthermore, the building survey object generation method further includes acquiring volume information of a building, and updating the survey object in the building based on the volume information.

[0041] This allows the building's survey targets to be updated, and further allows survey targets in the survey waiting system to be identified based on the relevant information of the updated survey targets, which is advantageous in improving the accuracy and efficiency of survey target generation and avoiding overlooking survey targets.

[0042] Furthermore, the method further includes obtaining survey target information of at least one survey target in the building and information on influencing factors that affect the appearance of the survey target, and identifying appearance information of the at least one survey target based on the survey target information and the influencing factor information.

[0043] As a result, by generating appearance information of the object to be investigated in a building based on the corresponding object information and influencing factor information, it is possible to obtain appearance information of the object to be investigated that matches the image of the investigation site, improving the accuracy of the generation of appearance information.In addition, the present invention can automatically generate appearance information of the object to be investigated, eliminating the need for professional engineers to manually select it, improving the efficiency of generating appearance information and saving on investigation costs.Furthermore, by using appearance information that matches the investigation site, it is advantageous for investigators to quickly find the object to be investigated, improving the efficiency of building investigations and saving on investigation costs.

[0044] Furthermore, the survey target information includes initial appearance information of the survey target, and in the method for generating appearance information of a building survey target according to an embodiment of the present invention, the step of "identifying appearance information of the at least one survey target based on the survey target information and the influencing factor information" includes inputting the initial appearance information of the survey target and the influencing factor information into a first generation model and outputting standard appearance information of the at least one survey target.

[0045] In this way, the "first generation model" is used to process the "initial appearance information" and influencing factor information in the information of the object to be surveyed, and standard appearance information is generated, eliminating the need to manually select the standard appearance information of the object to be surveyed, and further improving the efficiency and accuracy of generating appearance information. This further improves the efficiency of building surveys, reduces labor costs, and the generated model has generalizability, allowing it to generate standard appearance information that suits the on-site situation based on new, unseen data, with greater flexibility.

[0046] Furthermore, the first generative model includes a generative adversarial network (GAN) or a three-dimensional generative adversarial network (3D-GAN).

[0047] This increases the diversity of appearance information generated using the model, which is advantageous for meeting the needs of different usage scenarios.

[0048] Furthermore, the first generative model is trained based on at least initial appearance information of each type of survey object in the building, survey data generated according to the needs of each survey, and standard appearance information of the survey object corresponding to the survey data, and the survey data includes information on influencing factors of the survey object.

[0049] This allows the first generative model to be trained using at least the "initial appearance information of the survey object," "survey data," and "standard appearance information corresponding to the survey data," thereby further improving the efficiency and accuracy of generating standard appearance information of the survey object, and further improving the efficiency of building surveys.

[0050] The method further includes training the first generative model, wherein the step of training the first generative model includes obtaining training samples including initial appearance information of various types of survey objects in the building, corresponding survey data generated according to the needs of each survey, and standard appearance information of the survey objects corresponding to the survey data, and training an initial first generative model using the training samples, wherein the standard appearance information output from the first generative model is compared with the standard appearance information previously obtained, and the trained first generative model is obtained so that a loss function of the first generative model is minimized.

[0051] This improves the accuracy of the generative model, and further improves the efficiency and accuracy of generating appearance information of the object of investigation, further improving investigation efficiency.

[0052] Further, obtaining the initial appearance information of the object of investigation includes searching for corresponding appearance information as the initial appearance information of the object of investigation from a database based on the identification information of the object of investigation, or generating the initial appearance information of the object of investigation based on the identification information of the object of investigation and a second generative model.

[0053] This eliminates the need to manually select the initial appearance information of the object of investigation, and improves the efficiency and accuracy of obtaining the initial appearance information, thereby improving the efficiency and accuracy of generating standard appearance information of the object of investigation and further improving investigation efficiency.

[0054] Furthermore, the survey target information includes initial appearance information of the survey target and each of a plurality of related survey targets related to the survey target, and identifying the appearance information of the at least one survey target based on the survey target information and the influencing factor information includes obtaining initial appearance information of the survey target and each of a plurality of related survey targets related to the survey target, and generating standard appearance information including the plurality of survey targets based on the initial appearance information of the survey target and each of a plurality of related survey targets related to the survey target and the influencing factor information.

[0055] This eliminates the need to manually select standard appearance information for the object of investigation, is simple to implement, and allows standard appearance information for the object of investigation to be obtained quickly, improving the efficiency and accuracy of generating appearance information, thereby improving the efficiency of building investigations and reducing labor costs.

[0056] Furthermore, generating standard appearance information including the multiple survey objects based on the appearance information of the survey object and multiple related survey objects related to the survey object and the influencing factor information includes combining initial appearance information of the survey object and multiple related survey objects related to the survey object based on the influencing factor information, and treating the combined appearance information as standard appearance information including the multiple survey objects.

[0057] This further improves the efficiency and accuracy of appearance information generation, thereby further improving the efficiency of building inspection. Furthermore, compared to the form in which standard appearance information is identified using a generative model, this embodiment does not require training a model, reducing the cost required for training a model and the difficulty of implementation.

[0058] Furthermore, the appearance information is represented by at least one of a picture, a video, a movie, and a three-dimensional model.

[0059] This makes the present application applicable to the display of various types of research objects, meeting different display needs, and having a wide range of applicability and versatility.

[0060] Furthermore, the survey target includes at least one of architectural space within the building, building materials, ducts, wiring, equipment within the building, and the mounting or placement space of the equipment.

[0061] This makes the present invention applicable to various types of survey targets within a building, making it highly versatile and applicable in a wide range of applications.

[0062] Furthermore, the impact factor information includes at least one of building information, the type of subsystem in the building, the number of subsystems in the building, the type of equipment included in the subsystem in the building, and the number of equipment included in the subsystem in the building.

[0063] As a result, the present application takes into account various types of influencing factors within a building when generating appearance information, making it applicable in a wide range of scenarios and highly versatile.

[0064] Furthermore, the building information includes at least one of the number, location, spatial size, building materials, and building volume information of the placement spaces of similar equipment included in the building's subsystems, and the subsystems in the building and the systems awaiting investigation each include at least one of the building's energy system, HVAC centralized control system, air management system, lighting system, security system, architectural model system, conference system, entry system, and guest system.

[0065] This makes the present application applicable to various types of buildings and subsystems, and has a wide range of application and high versatility.

[0066] Furthermore, the impact factor information includes at least one of building information, the type of subsystem in the building, the number of subsystems in the building, the type of equipment included in the subsystem in the building, and the number of equipment included in the subsystem in the building.

[0067] As a result, the present application takes into account various types of influencing factors within a building when generating appearance information, making it applicable in a wide range of scenarios and highly versatile.

[0068] Furthermore, the influencing factor information is obtained from data obtained by surveying users.

[0069] This can improve the reliability and accuracy of the influencing factor information, thereby further improving the efficiency and accuracy of generating appearance information of the investigation object.

[0070] Furthermore, with regard to the building inspection method according to the second aspect of the present invention, the step of "inspecting the generated inspection targets" includes transmitting the generated inspection targets to a terminal device, receiving field data corresponding to the inspection targets transmitted from the terminal device, obtaining standard data corresponding to each of the inspection targets, and determining whether each inspection target meets the standard based on the field data and the standard data.

[0071] This allows researchers to quickly obtain survey subjects via terminal equipment and upload field data corresponding to each survey subject in a timely manner, which is advantageous in improving the efficiency of field data collection and further in improving survey efficiency.

[0072] Furthermore, the building survey method further includes generating appearance information of at least one survey object to be surveyed based on a building survey object generation method relating to a first aspect of an embodiment of the present invention, and surveying the at least one survey object based on the appearance information.

[0073] This allows the efficiency and accuracy of building inspections to be improved by investigating the target based on appearance information, thereby reducing inspection costs.

[0074] Furthermore, "inspecting the at least one survey object based on the appearance information" includes transmitting appearance information of the at least one survey object to a terminal equipment, receiving site data corresponding to the at least one survey object transmitted from the terminal equipment, and determining whether the at least one survey object conforms to a standard based on the appearance information and the corresponding standard data.

[0075] This allows investigators to obtain appearance information of the survey targets through terminal equipment, allowing them to quickly find the survey targets based on the appearance information and upload field data corresponding to each survey target in a timely manner, which is advantageous for improving the efficiency of field data collection and further for improving survey efficiency.

[0076] Furthermore, the on-site data, the standard data, and the appearance information include image data or three-dimensional model data.

[0077] This allows the data type, display format and processing format of the on-site data, standard data and appearance information to be selected as needed, which is advantageous in improving the efficiency of data matching and further improving the efficiency of investigating the survey object.

[0078] With reference to the following description and drawings, specific embodiments of the present invention are disclosed in detail, demonstrating the manner in which the principles of the present invention may be employed. The embodiments of the present invention should not be considered restrictive in scope. Many variations, modifications, and equivalents are encompassed within the spirit and terms of the appended claims.

[0079] Feature information described and shown in one embodiment may be used in the same or similar manner in one or more other embodiments, combined with feature information in the other embodiments, or substituted for feature information in the other embodiments.

[0080] It must be emphasized that the term "comprises" when used in the present text refers to the presence of a feature, an entire element, a step or an element, but does not exclude the presence / addition of one or more other features, entire elements, steps or elements.

[0081] The above and other objects, features, and advantages of the embodiments of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings.

[0014] Figure 1 is a block diagram of a method for generating building survey objects according to an embodiment of the present invention.

[0015] Figure 2 is a schematic diagram of steps for generating survey objects according to an embodiment of the present invention.

[0016] Figure 3 is a schematic diagram of steps for identifying systems to be surveyed according to an embodiment of the present invention.

[0017] Figure 4 is a schematic diagram of a method for training a survey object generation and deduplication model according to an embodiment of the present invention.

[0018] Figure 5 is a schematic diagram of the data in Table 2 after visualization.

[0019] Figure 6 is a schematic diagram of the data in Table 3 after visualization.

[0020] Figure 7 is a flowchart for generating survey objects according to an embodiment of the present invention.

[0021] Figure 8 is a schematic diagram of other steps for generating survey objects according to an embodiment of the present invention.

[0022] Figure 9 is a schematic diagram of steps for generating and using a lookup table according to an embodiment of the present invention.

[0023] Figure 10 is a schematic diagram of a survey object graph data structure before update according to an embodiment of the present invention.

[0024] Figure 11 is a schematic diagram of a survey object graph data structure after update according to an embodiment of the present invention.

[0025] Figure 12 is a block diagram of a method for generating appearance information of building survey objects according to an embodiment of the present invention.

[0026] Figure 13 is a schematic diagram of steps for identifying appearance information of a survey object according to an embodiment of the present invention.

[0027] Figure 14 is a schematic diagram of generating appearance information of a survey object using a generative adversarial network (GAN) according to an embodiment of the present invention.

[0028] Figure 15 is a schematic diagram of training a first generative model including a generative adversarial network (GAN). FIG. 1 is a schematic diagram of training a first generative model including a three-dimensional generative adversarial network (3D-GAN); FIG. 2 is a schematic diagram of a process of searching for initial appearance information from a database according to an embodiment of the present invention; FIG. 3 is a schematic diagram of another step of identifying appearance information of a building object to be inspected according to an embodiment of the present invention; FIG. 4 is a schematic diagram of generating standard appearance information using mapping according to an embodiment of the present invention; FIG. 5 is a block diagram of a building inspection method according to an embodiment of the present invention; FIG. 6 is a schematic diagram of a step of generating appearance information of a building object to be inspected according to an embodiment of the present invention; and FIG. 7 is a block diagram of building inspection equipment according to an embodiment of the present invention.

[0082] These and other features of the present invention will become apparent from the following specification, which refers to the drawings, in which: While the specification and drawings specifically disclose certain embodiments of the invention and illustrate some of the embodiments in which the principles of the invention may be employed, it is to be understood that the invention is not limited to the described embodiments, but includes all modifications, variations, and equivalents that fall within the scope of the appended claims.

[0083] Hereinafter, a building survey object generation method, a building survey method, a building survey facility, and a system according to embodiments of the present application will be described with reference to the drawings.

[0084] An embodiment of the first aspect of the present invention provides a building survey object generation method, and Fig. 1 is a block diagram of the building survey object generation method according to the embodiment of the present invention. As shown in Fig. 1, the building survey object generation method 100 includes step 101 of generating, for at least one system to be surveyed in a building, a plurality of survey objects corresponding to the at least one system to be surveyed in a deduplication manner, wherein the generated plurality of survey objects are different from each other, and the deduplication manner is implemented based on at least the correspondence between the systems to be surveyed and the survey objects.

[0085] This allows for the automatic generation of multiple, different survey targets for systems in a building awaiting investigation using a duplication elimination method, thereby solving the problem of overlapping and redundant survey targets, which is advantageous for improving survey efficiency and reducing survey costs.Furthermore, by generating survey targets for at least one system awaiting investigation, in cases where it is necessary to conduct surveys on multiple systems awaiting investigation simultaneously and related or identical survey targets exist in multiple systems awaiting investigation, it is possible to generate survey targets for all of the multiple systems awaiting investigation that do not overlap with each other.Furthermore, investigators can carry out survey work on multiple systems awaiting investigation at once and do not need to repeatedly survey any of the survey targets, which is advantageous for further improving survey efficiency.

[0086] In some embodiments, the building may be any type of building, such as an office building, a shopping mall, an industrial plant, a school, an apartment building, a residential house, etc. The embodiments of the present invention do not limit the type of building.

[0087] In some embodiments, the inspection waiting system may be various systems installed in a building, including, for example, a building energy system, an HVAC (Heating Ventilation and Air Conditioning) central control system, an air management system, a lighting system, a security system, an architectural model system, a conference system, an entry system, a guest system, etc. The embodiments of the present invention do not limit the type of inspection waiting system in a building.

[0088] In some embodiments, the survey target may be various objects that need to be surveyed in a building, and the survey target may include, for example, at least one of architectural spaces (rooms) in the building, building materials, conduits, wiring (e.g., the layout of lines such as Internet lines and KNX lines), equipment in the building (e.g., equipment such as cabinets, control panels, lamps, and switches), and the installation or placement space of the equipment. The embodiments of the present invention do not limit the type of survey target. Here, the architectural spaces (rooms) may include at least one of visible architectural spaces and hidden architectural spaces. The architectural spaces may include, for example, the size of the architectural space, and the installation or placement space of the equipment may include, for example, the size of the installation or placement space of the equipment.

[0089] Additionally, in some embodiments, the survey objects may be referred to as survey points.

[0090] In some embodiments, there is a correspondence relationship between the system to be investigated and the investigation targets, such as a "one-to-many" relationship, i.e., one system to be investigated corresponds to multiple investigation targets. An investigation of the system to be investigated is effectively an investigation of all investigation targets corresponding to the system to be investigated. Therefore, when it is necessary to investigate the system to be investigated, first, the investigation targets corresponding to the system to be investigated can be identified, and then the identified investigation targets can be investigated.

[0091] In some embodiments, one investigation target corresponds to at least one (one or more) systems to be investigated, i.e., two or more systems to be investigated may all have a correspondence relationship with the same investigation target. For example, system to be investigated A corresponds to investigation targets 1, 2, and 3, system to be investigated B corresponds to investigation targets 1 and 4, and investigation target 1 corresponds to systems to be investigated A and B, respectively.

[0092] When one investigation task simultaneously includes investigations of system A and system B awaiting investigation, according to conventional methods, it is necessary to investigate investigation targets 1, 2, and 3 for system A awaiting investigation, and to investigate investigation targets 1 and 4 for system B awaiting investigation. This results in duplicate investigation of investigation target 1, which affects investigation efficiency.

[0093] In an embodiment of the present invention, in step 101, for at least one system awaiting investigation, a plurality of investigation targets corresponding to the at least one system awaiting investigation are generated using a deduplication method, and the generated plurality of investigation targets are each different. For example, for system A awaiting investigation and system B awaiting investigation, a plurality of investigation targets corresponding to the entire system A awaiting investigation and system B awaiting investigation are generated using a deduplication method, namely investigation target 1, investigation target 2, investigation target 3, and investigation target 4. Since the investigation targets generated using the deduplication method are each different and do not overlap, duplicate investigations of the same investigation target can be avoided, which is advantageous in improving investigation efficiency.

[0094] The above has described the survey target, the survey waiting system, and the relationship between them. Hereinafter, the survey target generation method will be specifically described using an example.

[0095] In some embodiments, the search object can be generated by a trained model.

[0096] 2 is a schematic diagram of the step of generating investigation objects according to an embodiment of the present invention. As shown in FIG. 2, the step 101 of generating a plurality of investigation objects corresponding to the at least one system to be investigated using a deduplication method includes the step 1011 of outputting a plurality of investigation objects corresponding to the at least one system to be investigated using a pre-trained investigation object generation and deduplication model.

[0097] This allows for the efficiency and accuracy of generating survey targets to be improved and the efficiency of building surveys to be improved by identifying survey targets using the survey target generation and deduplication model.

[0098] Furthermore, the "survey target generation / duplication elimination model" may be named differently as long as it has the function of "generating a survey target using a duplication elimination method for at least one system awaiting investigation," and embodiments of the present invention do not limit the name of the model.

[0099] In some embodiments, as shown in FIG. 2, when "generating multiple research objects using a deduplication method" means "outputting multiple research objects using a research object generation and deduplication model," step 101 further includes step 1012 of training the research object generation and deduplication model.

[0100] Here, the step of training the search object generation and deduplication model includes training an initial model using a training sample obtained in advance to obtain the search object generation and deduplication model.

[0101] Since the search target generation and deduplication model is generated in advance, step 1012 may be performed independently or may be performed in advance. That is, step 101 does not necessarily include step 1012. Therefore, in Fig. 2, step 1012 is indicated by a dashed box as an optional step, and may be omitted in the actual process of identifying the search target. Furthermore, Fig. 2 may further include other steps, and the present invention is not limited thereto.

[0102] In some embodiments, "initial model" refers to an untrained subject generation and deduplication model.

[0103] In some embodiments, the initial model is a neural network model, for example a graph neural network model, and accordingly, the search object generation and deduplication model is also a graph neural network model.

[0104] In some embodiments, the survey target generation and deduplication model is trained based on at least the correspondence between systems to be surveyed and survey targets and pre-set survey target deduplication rules.

[0105] Accordingly, the training sample for training the survey target generation and deduplication model includes at least a system to be surveyed, a survey target, a correspondence relationship between the system to be surveyed and the survey target, and a predetermined survey target deduplication rule.

[0106] In this case, the survey target generation / duplication elimination model learns in advance the correspondence between the systems to be surveyed and the survey targets, as well as the preset survey target duplication elimination rules, so that when at least one system to be surveyed is input as input data into the survey target generation / duplication elimination model, the survey target generation / duplication elimination model outputs multiple survey targets corresponding to the at least one system to be surveyed, and these multiple survey targets are each different from one another.

[0107] In some embodiments, the systems awaiting investigation, the survey targets, and the correspondence between the systems awaiting investigation and the survey targets are stored in advance. For example, the systems awaiting investigation, the survey targets, and the correspondence between the systems awaiting investigation and the survey targets in a building are collected and stored by relevant parties, or the systems awaiting investigation, the survey targets, and the correspondence between the systems awaiting investigation and the survey targets in a building are automatically generated and stored by a computer system based on the relationships between the systems awaiting investigation, the survey targets, and the systems awaiting investigation and the survey targets in a building.

[0108] In some embodiments, the search object de-duplication rules are pre-defined, for example, defined and stored by a participant, and the search object de-duplication rules are used to deduplicate the search objects.

[0109] In some embodiments, the survey target generation and deduplication model is trained based on the correspondence between systems to be surveyed and survey targets, pre-defined survey target deduplication rules, and pre-defined association relationships between survey targets.

[0110] Accordingly, the training sample for training the survey target generation and deduplication model includes at least a system to be surveyed, the survey target, a correspondence relationship between the system to be surveyed and the survey target, a predetermined survey target deduplication rule, and a predetermined association relationship between the survey targets.

[0111] In addition, step 101 of generating a plurality of survey objects corresponding to the at least one system to be surveyed using a deduplication method includes outputting the plurality of survey objects and the association relationships between the plurality of survey objects using the survey object generation / deduplication model.

[0112] At this time, the survey target generation / duplication elimination model has previously learned the correspondence between the systems to be surveyed and the survey targets, the preset survey target duplication elimination rules, and the preset association relationships between the survey targets, and when at least one system to be surveyed is input as input data into the survey target generation / duplication elimination model, the survey target generation / duplication elimination model outputs multiple survey targets corresponding to the at least one system to be surveyed and the association relationships between these multiple survey targets, and these multiple survey targets are each different from one another.

[0113] As a result, the survey target generation and duplication elimination model outputs not only the survey targets but also the relationships between each survey target, which helps researchers to focus their research on multiple related survey targets, such as different survey targets in the same location, and avoids researchers having to repeatedly travel to the same location, which is advantageous for improving survey efficiency.

[0114] In some embodiments, the association relationship between the study objects comprises at least one of an electrical relationship, a mechanical relationship, and a spatial relationship between the study objects.

[0115] This helps the researcher quickly understand the electrical, mechanical, and spatial relationships between each of the survey objects that require investigation, which is advantageous for understanding the relationships between survey objects early on and improving survey efficiency.

[0116] Here, electrical relationships include, for example, relationships between electrical energy production, electrical energy transmission, electrical energy distribution, and electrical energy detection. For example, an energy consumption system is a system that monitors electrical energy usage. The investigation target varies depending on the user's level of segmentation of the monitored subsystems. If the user's needs are the total power consumption of the monitored subsystem, the investigation target is the total circuit of the corresponding subsystem. If the user's needs are the power consumption of a subsystem in the monitored subsystem, the investigation target is a single circuit of that subsystem. For example, as shown in Figure 5, if the investigation target is low-voltage room 1, circuit 1, and lamp 1, and lamp 1 is electrically connected to low-voltage room 1 via circuit 1, the electrical relationship between low-voltage room 1, circuit 1, and lamp 1 can also be generated when generating the investigation target. This allows the investigator to easily understand the electrical relationships between the investigation targets.

[0117] The mechanical relationships include, for example, mechanical structures and mechanical connection relationships. For example, a building or structure may include multiple subsystems, and all of the subsystems may be ultimately connected to form a local area network. The local area network may be constructed by a network switch, and the network switch may be mechanically attached to an IT machine room. That is, the network switch and the IT machine room may be mechanically connected. Therefore, a research object corresponding to the IT machine room may be generated by a subsystem corresponding to the user's needs, and a relationship between the IT machine room and the network switch may be generated.

[0118] The spatial relationship includes, for example, spatial parallelism, overlap, and inclusion. For example, when the building or structure being surveyed is large, multiple IT cabinets are distributed in different areas of the building due to limitations imposed by the length of subsystem communication lines. All distributed IT cabinets are grouped together in an IT machine room, and some of the subsystem survey objects are included in the IT machine room. Some of the survey objects included in the IT machine room overlap with the survey objects of the IT cabinet. Therefore, the overlapping survey objects are removed, leaving only the survey objects in the IT machine room. A spatial inclusion relationship (association relationship) can then be generated between the survey objects in the IT machine room, IT cabinet, and IT cabinet. Here, the IT machine room, IT cabinet, and components or devices in the IT cabinet belong to the survey objects of the first, second, and third levels, respectively.

[0119] As described above, different relationships may be distinguished from one another by different representations, or different relationships may be de-duplicated, for example, only one connecting line may be displayed between two related research objects. Specifically, different relationships may be displayed in different images or in the same image (e.g., FIG. 5). Furthermore, different colors of connecting lines may represent different relationships or different types of relationships, and this application is not limited thereto.

[0120] In some embodiments, the association relationship between the investigation objects is stored in advance (e.g., stored in a database) and used to identify related investigation objects related to the investigation object. For example, the association relationship between "power supply panel" and "air switch" is stored in advance in the database, and if the investigation object includes "power supply panel", it can be determined based on the association relationship that the related investigation object of "power supply panel" includes "air switch", or it can be directly determined that "air switch" is located in "power supply panel".

[0121] The pre-stored association relationships can be used in the "model method" and "mapping" in subsequent embodiments of the present application. In the "model method," the pre-stored association relationships are trained in a model during the model training stage, so that the trained model can output information including the investigation target and related investigation targets as standard appearance information of the investigation target based on the input initial appearance information and influencing factor information of the investigation target. In the "mapping," related investigation targets related to the investigation target can be identified by querying a database. For details, please refer to the related embodiments of the present invention and will not be described in detail here.

[0122] In some embodiments, as shown in FIG. 2, when the input data of the research target generation and deduplication model is a system to be researched, step 101 further includes step 1013 of identifying the at least one system to be researched according to user needs.

[0123] Here, the systems to be investigated are obtained in response to the user's needs. The user's needs may take the form of a needs table, for example. The needs table may include all systems to be investigated in a building and markers for the systems to be investigated by the user. The markers are used to indicate the user's needs for the systems to be investigated. The creation needs refer to the need for system creation for a new building, where a system has not yet been installed. The modification needs refer to the need for system modification for an existing building, where a system has already been installed but the system needs to be upgraded due to low intelligence or other reasons. For example, if a user wants to create or modify one or more of the systems to be investigated, the user may mark positions corresponding to the one or more systems to be investigated as "needing creation or modification" or mark positions corresponding to systems to be investigated other than the one or more systems to be investigated as "not requiring creation or modification." The markers may be, for example, pre-set characters, which may include, for example, symbols, alphabets, numbers, Chinese characters, or a combination thereof. The "markers requiring creation or modification" and the "markers not requiring creation or modification" correspond to different characters.

[0124] The needs table may be an electronic file, for example, an electronic file created directly by the user or an electronic file obtained by scanning a paper file marked by the user, and the present application is not limited thereto.

[0125] In Fig. 2, step 1013 is indicated by a dashed box as an optional step, and may be omitted in the actual process of identifying the research target. In addition, Fig. 2 may further include other steps, and the present invention is not limited thereto.

[0126] 3 is a schematic diagram of a step of identifying a system to be investigated according to an embodiment of the present invention. As shown in FIG. 3, step 1013 of identifying the at least one system to be investigated according to user needs includes step 1013-1 of acquiring a user's needs table including a plurality of systems to be investigated and user's markers for the systems to be investigated, and step 1013-2 of identifying the systems to be investigated marked by the user as the at least one system to be investigated.

[0127] This allows systems to be identified based on the user's needs, and then identifies research targets based on the identified systems to be investigated, eliminating the need to manually convert the user's needs into systems to be investigated, which is advantageous in improving the efficiency of generating research targets and further improving research efficiency.

[0128] In the above example, the "marker" is a "marker that requires investigation," and if the user adopts a "marker that does not require investigation" in the needs table, in step 1013-2, a system awaiting investigation that is not marked by the user is identified as the at least one system awaiting investigation.

[0129] In some embodiments, the types of the systems to be investigated include a first investigation system and a second investigation system, and the first investigation system includes one or more second investigation systems, for example, in an energy system, it includes an energy consumption billing system, a water meter collection data visualization system, an inquiry and report system, etc. Here, the second investigation system may be referred to as a function point under the first investigation system, for example, energy consumption billing and data visualization are function points under the energy system.

[0130] The needs table may include a first system to be investigated and a second system to be investigated (i.e., function points), and the user may mark the second system to be investigated and identify the first system to be investigated and the corresponding second system to be investigated marked by the user as the at least one system to be investigated. For example, an energy system - an energy consumption billing system may be identified as one system to be investigated.

[0131] In some embodiments, the search object generation and de-duplication model is trained based on at least a correspondence between user needs and search objects and a preset search object de-duplication rule.

[0132] Accordingly, the training sample for training the search target generation and deduplication model includes at least user needs, search targets, correspondence between the user needs and the search targets, and pre-set search target deduplication rules.

[0133] In this case, the survey target generation / duplication elimination model learns in advance the correspondence between user needs and survey targets and the preset survey target duplication elimination rules, so that when the user needs are input as input data into the survey target generation / duplication elimination model, the survey target generation / duplication elimination model outputs multiple survey targets that correspond to the user needs, and these multiple survey targets are each different from one another.

[0134] Regarding the content relating to user needs, the correspondence between user needs and research targets, and research target duplication elimination rules, please refer to the above embodiment and the description will not be repeated here.

[0135] In some embodiments, the search object generation and de-duplication model is trained based on the correspondence between user needs and search objects, pre-defined search object de-duplication rules, and pre-defined association relationships between search objects.

[0136] Accordingly, the training sample for training the survey object generation and deduplication model includes at least user needs, survey objects, correspondence relationships between user needs and survey objects, pre-set survey object deduplication rules, and pre-set association relationships between survey objects.

[0137] In addition, step 101 of generating a plurality of survey objects corresponding to the at least one system to be surveyed using a deduplication method includes outputting the plurality of survey objects and the association relationships between the plurality of survey objects using the survey object generation / deduplication model.

[0138] At this time, the survey object generation / duplication elimination model has previously learned the correspondence between user needs and survey objects, the preset survey object duplication elimination rules, and the preset association relationships between the survey objects. When the user's needs are input as input data into the survey object generation / duplication elimination model, the survey object generation / duplication elimination model outputs multiple survey objects that correspond to the user's needs and the association relationships between these multiple survey objects, and these multiple survey objects are each different from one another.

[0139] As a result, the survey target generation and duplication elimination model outputs survey targets and the relationships between each survey target based on the user's needs, helping researchers to focus their research on multiple survey targets that are related to each other, such as different survey targets in the same location, thereby avoiding repeated travel to the same location and improving survey efficiency.

[0140] In addition, this embodiment does not require the step of identifying at least one system to be investigated based on the user's needs shown in Figure 2 to be performed independently, but instead inputs the user's needs directly into the investigation target generation / duplication elimination model, which is advantageous in saving flow and improving the efficiency of identifying investigation targets and the efficiency of investigation.

[0141] In some embodiments, when user needs are used as input data for the research target generation and deduplication model, the method further includes formatting the needs table to obtain graph data corresponding to the at least one system to be researched.

[0142] This allows the user needs of the graph data format to be directly input into the search object generation and deduplication model trained based on the graph neural network to obtain the corresponding search objects, which is advantageous for improving the efficiency of search object generation and further improving the search efficiency.

[0143] In some embodiments, the building survey object generation method further includes determining a survey ranking of the plurality of survey objects based on association relationships between the survey objects.

[0144] This allows for a rational arrangement of the survey order, and provides researchers with a reference for the order in which to survey each survey subject, which is advantageous for improving survey efficiency.

[0145] In some embodiments, a search object generation and de-duplication model is utilized to determine a search order for the plurality of search objects.

[0146] Accordingly, the training samples for training the survey object generation and deduplication model also include the survey rankings of the survey objects in the building, so that the survey object generation and deduplication model has previously learned the survey rankings of the survey objects in the building, and when the survey waiting system or user needs are input as input data to the survey object generation and deduplication model, the survey object generation and deduplication model outputs the corresponding multiple survey objects and the survey rankings of the multiple survey objects.

[0147] In some embodiments, the search order of a search object includes, for example, a search priority of the search object.

[0148] In some embodiments, the search order of the search objects can be determined based on the relationship between the search objects. This makes the search order generated based on the relationship between the search objects more rational, and directly generating the search order provides more intuitive instructions to the researcher. The search order can be visualized in the form of a sequence code (e.g., a number), but is not limited thereto. For example, as shown in FIG. 5, the search order can be inferred as follows: (1) IT machine room → (2) IT cabinet → (3) network electrical box 1, network electrical box 2, network electrical box 3 → (4) monitoring host → ...

[0149] The following describes the process of training the search object generation and deduplication model using an example. In this example, the input data of the trained search object generation and deduplication model are user needs, and the output data is a plurality of search objects with a search order. Furthermore, the training process of the search object generation and deduplication model when the input data is a system to be searched mentioned in the above example and the training process of the search object generation and deduplication model when the output data is a plurality of search objects or search objects with associated relationships are similar, but the main difference is the training samples, so you can refer to them and do it yourself, and we will not repeat the description here.

[0150] FIG. 4 is a schematic diagram of a method for training a survey target generation and deduplication model according to an embodiment of the present application. As shown in FIG. 4, the training method, i.e., the step of training a survey target generation and deduplication model, includes S11 of acquiring training samples including multiple pieces of training data, each of which includes "input data" and "output data", where the "input data" is "user needs" or "systems waiting to be surveyed", and the output data is S11 of "survey targets" corresponding to the "input data", and / or "association relationships between the survey targets", and / or "survey rankings of the survey targets", S12 of constructing a graph data structure conversion model, S13 of constructing a graph neural network, S14 of determining a loss function, and S15 of training using the training samples until the loss function of the model satisfies a target value.

[0151] Here, in S11, the "input data" of the training data is user needs, for example, a user needs table, and Table 1 is an example of a user needs table. The data in Table 1 can be visualized, for example, by a graph data structure transformation model in step S12, and a graph data structure corresponding to the "input data" is obtained, which is then used to train the graph neural network in S13.

[0152] The "output data" of the training data is data of survey targets with survey rankings, and Table 2 is an example of data of survey targets with survey rankings.

[0153] In Table 2, the left column represents the survey objects, each of which appears as a triplet. For example, the triplet (0, 'room class', 'IT machine room') in Table 2 contains the survey object ID "0", the survey object type "room class", and the survey object name "IT machine room", and the meanings of the remaining triplets are similar. The right column represents the survey order of the survey objects, each of which appears as a tuple. For example, the tuple (0, 1) in Table 2 contains the source survey object ID "0" and the target survey object ID "1", in that order. That is, the survey object with ID "0" is surveyed first, and then the survey object with ID "1", and the meanings of the remaining tuples are similar.

[0154] The data in Table 2 can be visualized, and Fig. 5 is a schematic diagram of the visualized data in Table 2. As shown in Fig. 5, the associations and search rankings of the search targets are represented by a graph data structure.

[0155] In step S11, graph data of a needs table of historical user needs may be used as "input data" of the training data, and data obtained by an expert optimizing the ranking of survey objects corresponding to the needs table of the history may be converted into survey ranking graph data and used as "output data." Alternatively, a needs table may be randomly generated and converted into graph data and used as "input data," and survey rankings may be determined by an expert based on the randomly generated needs table, and then converted into survey ranking graph data and used as "output data."

[0156] In S12, the graph data structure conversion model is used to convert the function points (i.e., minimum demand points) under the marked subsystems in the two-dimensional needs table into a graph data structure to be used as "input data" in the training sample. For example, the function points under the subsystems marked by the user in Table 1 are converted into the graph data structure shown in Figure 5. Hereinafter, another example is provided, assuming that the "input data" is the user's needs, including the user's function points and the IDs, main systems, and subsystems corresponding to the function points. Table 3 is an example of the "input data". In Table 3, the input data includes multi-stage data, but in actual application, the input data may only include one stage.

[0157] Taking the function point "Water Meter Collection" as an example, its corresponding subsystem is the "Visualization System" and its corresponding main system is the "Energy Subsystem." Based on the dependency relationships among function points, subsystems, and main systems, Table 3 can be converted into a graph data structure using a graph data structure conversion model, and Figure 6 is a schematic diagram of the data in Table 3 after it has been visualized. As shown in Figure 6, the graph data structure represents the system to be investigated and the user's needs.

[0158] The step of generating a graph data structure according to the graph data structure transformation model includes the following steps.

[0159] S12.1: Go through all function points.

[0160] S12.2: For each function point, search for its upper level point (e.g., subsystem), and if the upper level point cannot be found, create this point and connect the function point to the created upper level point, and if the upper level point is found (already created), connect the function point to the found (already created) upper level point.

[0161] S12.3: For the upper point, the point two levels above it (i.e., the points two levels above the function point, e.g., the main system) is searched for, and if the upper point cannot be found, this point is created and the upper point is connected to the created upper point; if the upper point is found (already created), the upper point is connected to the found (already created) upper point.

[0162] S12.4: Repeat S12.2 and S12.3 until all function points have been cycled through.

[0163] In S13, the steps of constructing a graph neural network are as follows:

[0164] S13.1: Based on the "input data", construct the corresponding types, point features, and edge features and load them using HeterGraphWrapper.

[0165] S13.2: Based on the "output data", construct the corresponding types, point features, and edge features and load them using HeterGraphWrapper.

[0166] S13.3: The loaded "input data" is mapped to the loaded "output data" by the message_passing function (whose main role is to calculate features), by a two-layer fully connected neural network layer, and then by a softmax layer.

[0167] This completes the construction of a graph neural network.

[0168] In S14, based on the graph neural network model and training data, sigmoid_cross_entropy_with_logits can be used as the loss function, for example, the formula is as follows:

[0169] In S15, when the model's loss function is trained using the training samples until it meets the target value, the present research object generation and deduplication model is obtained. This research object generation and deduplication model can output a research object ranking diagram or can directly output the research objects. Here, the research object ranking diagram can be of various types and is a heterogeneous graph.

[0170] Accordingly, FIG. 7 is a flowchart of generating search objects according to an embodiment of the present invention, which utilizes a trained search object generation and deduplication model to convert user needs into a corresponding search object diagram.

[0171] Specific steps include, for example, the following steps:

[0172] Step 1: A step of acquiring user needs. A needs table containing each subsystem in the building and the function points under the subsystem is provided to the user. The customer marks one sub-function or function point in the needs table to acquire one sub-need. Table 4 is an example of a sub-need marked by the user. The marked sub-needs are combined to acquire the customer's user needs.

[0173] Step 2: A graph data structure is generated according to the user's needs, for example, the graph data structure shown in FIG.

[0174] Step 3: A step of generating a survey object. The graph data structure is input into a trained graph neural network (survey object generation / de-duplication model) and the survey object is output (the survey object appears as a survey object diagram, for example, the graph data structure shown in Figure 5, although the ID information in Figure 5 may be omitted).

[0175] In some embodiments, the search targets may be generated by a lookup table.

[0176] 8 is a schematic diagram of another step of generating survey objects according to an embodiment of the present invention. As shown in FIG. 8, step 101 of generating a plurality of survey objects corresponding to the at least one system to be investigated using a deduplication method includes: step 1014 of identifying a survey object corresponding to each system to be investigated in the at least one system to be investigated based on a pre-generated lookup table including at least a correspondence relationship between the system to be investigated and the survey objects; and step 1015 of deduplicating all the identified survey objects to obtain a plurality of survey objects corresponding to the at least one system to be investigated.

[0177] As a result, by using the lookup table to identify the survey target, an accurate survey target can be obtained quickly, improving the efficiency and accuracy of survey target generation and improving the efficiency of building surveys.

[0178] In the above embodiment, the research target is identified based on the system to be researched. Therefore, it is necessary to first identify the system to be researched according to the user's needs. For specific steps, please refer to the above embodiment and the description will not be repeated here.

[0179] In another embodiment, the research objects may be directly identified based on the user's needs, and in this case, there is no need to convert between the user's needs and the research waiting system. The specific steps are to identify multiple research objects corresponding to the user's needs based on a pre-generated lookup table, which includes at least the correspondence between the user's needs and the research objects, and then de-duplicate all the identified research objects to obtain multiple research objects that do not overlap with each other.

[0180] This eliminates the need to convert between the user's needs and the system waiting for investigation, which is advantageous in saving the flow and improving the efficiency of identifying the investigation target and the investigation efficiency.

[0181] In some embodiments, the lookup table further includes association relationships between the survey objects, and the building survey object generation method further includes identifying association relationships between the plurality of survey objects based on the lookup table.

[0182] This allows the researcher to use a lookup table to identify the research subjects and the relationships between them, which helps the researcher to focus on researching multiple research subjects that are related to each other, such as different research subjects in the same location, and avoids repeatedly traveling to the same location, which is advantageous in improving research efficiency.

[0183] In some embodiments, when a lookup table is used to identify the research target, step 101 further includes generating the lookup table. For example, the lookup table may be obtained by storing systems to be researched, research targets, and correspondences between the systems to be researched and the research targets, or the lookup table may be obtained by storing user needs, research targets, and correspondences between the user needs and the research targets.

[0184] In some embodiments, the building survey object generation method further includes determining a survey ranking of the plurality of survey objects based on association relationships between the survey objects.

[0185] This allows for a rational arrangement of the survey order, and provides researchers with a reference for the order in which to survey each survey subject, which is advantageous for improving survey efficiency.

[0186] In some embodiments, a lookup table is utilized to determine the search order of the plurality of search objects.

[0187] Accordingly, the lookup table further stores the inspection order of the inspection objects in the building, so that when the inspection waiting system or the user's needs are input data, the input data is checked against the lookup table to obtain the corresponding multiple inspection objects and the inspection order of the multiple inspection objects.

[0188] The following describes an example of the process of generating a lookup table and identifying research objects using the lookup table. In this example, the lookup table includes the correspondence between user needs and research objects and the association relationship between the research objects. In the process of identifying research objects using the lookup table, the input data is at least one of the system to be researched and the user needs, and the output data includes at least the research objects.

[0189] FIG. 9 is a schematic diagram of the steps of generating and using a lookup table according to an embodiment of the present application. As shown in FIG. 9, the step of training the survey target generation / de-duplication model includes the steps of: storing user needs, systems awaiting investigation, survey targets, correspondences between user needs and systems awaiting investigation, correspondences between systems awaiting investigation and survey targets, and associations between the survey targets to form a lookup table in S21; acquiring input data in S22; comparing the input data with the stored lookup table to identify corresponding survey targets in S23; and deduplicating the identified survey targets and outputting the de-duplication survey targets in S24.

[0190] Here, in S21, the user needs are, for example, function points, and the system awaiting investigation corresponding to the user needs may be a multi-tier system, for example, the system awaiting investigation corresponding to each function point includes its directly dependent subsystem and the main system dependent on the subsystem, or the system awaiting investigation corresponding to the user needs may be a single-tier system, for example, one function point directly dependent on a main system, and each main system corresponds to a unique ID. The correspondence relationship between the user needs and the system awaiting investigation is, for example, as shown in Table 3.

[0191] Table 5 shows an example of the correspondence between input data and the survey target.

[0192] In Table 5, ID is a unique ID corresponding to the function point, main system and subsystem are the main system and subsystem corresponding to the function point, respectively, related system is the main system in Table 3 related to the function point, and related table ID is an ID corresponding to the main system in Table 3.

[0193] In S22, the input data includes at least the system to be investigated, which may be, for example, the main system in Table 3. The input data may also include other data such as subsystems, user needs, etc.

[0194] Regarding the acquisition form of user needs data, for example, a needs table including each subsystem in a building and the function points under the subsystem is provided to the user, and the customer marks one sub-function or function point in the needs table to acquire one sub-need, and the marked sub-needs are compiled to obtain the customer's user needs.

[0195] Table 4 above is an example of sub-needs marked by a user. Table 6 shows another example of user needs. In Tables 4 and 6, the user's sub-needs are represented by the corresponding subsystems and sub-functions.

[0196] In step S23, the input data is compared with the stored lookup table to identify the corresponding research target. Specific steps are as follows:

[0197] S23.1: Extract keywords in the input data. For example, the input data includes a system to be investigated. The extracted keywords may include at least one of "energy," "HVAC," "security," "lighting," etc.

[0198] S23.2: Based on the extracted keywords, identify the main systems that contain the same keywords in a look-up table (here, for example, Table 3 above).

[0199] S23.3: Based on the ID of the queried main system, poll related IDs in another table (here, for example, Table 5 above) and identify the investigation target based on the related ID. For example, identify the investigation target corresponding to the same related ID as the ID of the queried main system as the investigation target.

[0200] S23.4: Based on other data in the input data (e.g., subsystem, function point), ambiguously query the function point data in Table 3, identify the corresponding system awaiting investigation and its ID from Table 3, and then poll the related ID in Table 5 based on the ID of the identified system awaiting investigation, and identify the investigation target based on the related ID, for example, identify the investigation target corresponding to the same related ID as the ID of the queried main system as the investigation target.

[0201] If the input data only includes systems awaiting investigation, step S23.4 may be omitted.

[0202] In S24, the identified investigation targets are de-duplicated and the investigation targets after de-duplication are output. For example, de-duplication is performed based on the related systems. Table 7 shows an example of investigation targets that have not been de-duplicated, and Table 8 shows an example of investigation targets after de-duplication of Table 7.

[0203]

[0204] In some embodiments, the building survey object generation method further includes representing the plurality of survey objects as at least one of a picture, a video, a text, a map, or a three-dimensional model, such as a point cloud model, a BIM model, or the like.

[0205] This provides various display modes of the survey object according to actual needs, and makes it easier for the surveyor to see the survey object, which is advantageous in improving survey efficiency.

[0206] In some embodiments, the building survey object generation method further includes obtaining volume information of a building, where the volume information may refer to data representing the volume of the building's space or the size of the volume of another building (e.g., the number of floors, the planar area of ​​each floor, etc.), and updating the survey object in the building based on the volume information.

[0207] Research has found that the survey objects corresponding to the same survey-waiting system in buildings with different volumes may be different, e.g., the quantities may be different. In order to make the generated survey objects more aligned with the actual situation in the building, we utilize volume information to update the generated survey objects, thereby making the survey objects more accurate, realistic, and reliable.

[0208] Fig. 10 is a schematic diagram of an investigation target graph data structure before an update according to an embodiment of the present invention, and Fig. 11 is a schematic diagram of an investigation target graph data structure after an update according to an embodiment of the present invention. As shown in Fig. 10 and Fig. 11, for example, based on the volume information of the building, a switch 3 has been added to the IT cabinet in the IT machine room, and surveillance cameras 4 and 5 have been newly added. Therefore, in the investigation target graph data structure after the update shown in Fig. 11, a switch 3 related to the original switch 2, and surveillance cameras 4 and 5 related to switch 3 have been added.

[0209] Furthermore, for example, based on the volume information of the building, one low-voltage room is added, and control panel 3 and control panel 4 are installed in the newly added low-voltage room, an HVAC control system is installed in control panel 3, and a lighting control system is installed in control panel 4. Therefore, in the updated survey target graph data structure shown in Figure 11, the original low-voltage room (see Figure 10) is changed to low-voltage room 1, and low-voltage room 2, control panel 3 and control panel 4 related to low-voltage room 2, an HVAC control system related to control panel 3, and a lighting control system related to control panel 4 are added.

[0210] Furthermore, for example, two HVAC systems are added to the HVAC circuit based on the volume information of the building, and therefore, in the updated survey target graph data structure shown in FIG. 11 , HVAC system 4 and HVAC system 5 related to the HVAC circuit are added.

[0211] Furthermore, for example, one lighting distribution room is added based on the volume information of the building, and the newly added lighting distribution room is provided with lines 3 and 4, with lamps 10, 11, and 12 arranged on line 3, and lamps 13, 14, and 15 arranged on line 4. Therefore, in the updated survey target graph data structure shown in FIG. 11, the original lighting distribution room (see FIG. 10) is changed to lighting distribution room 1, and lighting distribution room 2, lines 3 and 4 related to lighting distribution room 2, lamps 10, 11, and 12 related to line 3, and lamps 13, 14, and 15 related to line 4 are added.

[0212] This allows all survey targets in the building to be updated, and furthermore, based on the relevant information of all the survey targets after the update, survey targets in the survey waiting system that meet the user's needs can be identified, which is advantageous in improving the accuracy and efficiency of survey target generation and avoiding overlooking survey targets.

[0213] In some embodiments, the building survey object generation method further comprises:

[0214] When identifying a survey target based on the survey target generation / duplication elimination model, the survey target generation / duplication elimination model is updated based on the updated survey target. For example, a new training sample is identified based on information such as the updated survey target, the correspondence between the updated survey target and the systems waiting to be surveyed, the correspondence between the updated survey target and the user needs, and the association relationships between the updated survey targets, and the like, and the survey target generation / duplication elimination model is trained based on the new training sample to obtain an updated target generation / duplication elimination model.

[0215] When identifying a survey target based on a lookup table, the lookup table is updated based on the updated survey target. For example, a new lookup table is generated and stored for information such as the updated survey target, the correspondence between the updated survey target and the systems waiting to be surveyed, the correspondence between the updated survey target and the user needs, and the association between the updated survey targets.

[0216] This allows the survey target, survey target generation / deduplication model, or lookup table to be updated based on the building volume information, which is advantageous in ensuring the accuracy and efficiency of survey target generation and avoiding overlooking survey targets.

[0217] According to the above embodiment, for systems awaiting investigation in a building, the present invention can solve the problem of overlapping and redundant investigation targets by automatically generating multiple different investigation targets using a duplicate elimination method, which is advantageous for improving investigation efficiency and reducing investigation costs.In addition, by generating an investigation target for at least one system awaiting investigation, it is necessary to investigate multiple systems awaiting investigation simultaneously, and in cases where multiple systems awaiting investigation have related or identical investigation targets, it is possible to generate investigation targets for the multiple systems awaiting investigation that do not overlap with each other.Furthermore, investigators can carry out investigation work for multiple systems awaiting investigation in one go, and there is no need to repeatedly investigate any of the investigation targets, which is advantageous for further improving investigation efficiency.

[0218] The above describes an embodiment of the present invention for generating an investigation object. In another embodiment, the present invention further provides a method for generating appearance information of a building investigation object. After an investigation object is identified based on the building investigation object generation method of the present invention, appearance information of the building investigation object can be generated based on the building investigation object appearance information generation method of the present invention. Using appearance information that matches the investigation site is advantageous for investigators to quickly find investigation objects to investigate, further improving the efficiency of building investigations and further reducing investigation costs.

[0219] The method for generating appearance information of a building to be surveyed according to the present invention will be specifically described below.

[0220] FIG. 12 is a block diagram of a method for generating appearance information of a building survey target according to an embodiment of the present invention.

[0221] As shown in Figure 12, the method 200 for generating appearance information of a building survey target includes step 201 of acquiring survey target information and influencing factor information that affects the appearance of at least one survey target in the building, and step 202 of identifying appearance information of at least one survey target based on the survey target information and the influencing factor information.

[0222] This allows appearance information of the object to be investigated in a building to be generated based on the corresponding object information and influencing factor information, thereby obtaining appearance information of the object to be investigated that matches the image of the investigation site, and improving the accuracy of the generation of the appearance information.In addition, the present invention can automatically generate appearance information of the object to be investigated, eliminating the need for professional engineers to manually select it, improving the efficiency of generating appearance information and saving on investigation costs.Furthermore, by using appearance information that matches the investigation site, it is advantageous for investigators to quickly find the object to be investigated, improving the efficiency of building investigations and saving on investigation costs.

[0223] In some embodiments, the "survey target" targeted by the method for generating appearance information of a building survey target according to an embodiment of the present invention is generated based on the "building survey target generation method" in the above-described embodiment of the present invention.

[0224] In some embodiments, the "survey object" targeted by the method for generating appearance information of a building survey object according to an embodiment of the present invention may be a pre-set object, a default object, or an object obtained from data obtained by surveying users, and the present application is not limited thereto. For example, in the process of surveying users, the survey object may be identified according to the user's modification needs for making the building intelligent. For example, if the user needs to modify the lighting system, the survey object may be identified to include lamps, controllers or electrical control panels corresponding to the lamps, etc.

[0225] In some embodiments, the survey target information of the survey target includes, for example, at least one of identification information of the survey target (e.g., name, code number, etc.), initial appearance information of the survey target, specification information of the survey target, initial appearance information of different specifications of the survey target, and initial appearance information of related survey targets related to the survey target. The name of the survey target is, for example, a distribution board (which may also be referred to as a high-voltage distribution board, a low-voltage distribution board, a power supply board, etc.), the code number of the survey target is unique, and each survey target corresponds to one and only one code number, and the code number may include one of numbers, letters, and alphabets, or a combination thereof, the initial appearance information of the survey target refers to information about the appearance of a single survey target, the specification information of the survey target refers to information about the size specification of a single survey target, the initial appearance information of different specifications of the survey target refers to corresponding initial appearance information when the same survey target has different size specifications, and the related survey targets related to the survey target refer to survey targets that are related to the survey target. For details regarding the "related relationship between survey targets," please refer to the preceding paragraph and will not be repeated here.

[0226] In some embodiments, the related relationship of the survey object is expressed by the building information (see subsequent embodiments) in the influencing factor information. That is, for the same type of survey object, if the building information is different, the related relationship between the survey object and other survey objects may be different, and there may also be differences in the related survey objects related to the survey object. The related relationship that varies depending on the building information can be used in the "model method" described in subsequent embodiments of the present application. In this case, the building information is input into the trained model as one of the influencing factor information, and the standard appearance information output by the model includes the related survey objects of the survey object corresponding to the building information.

[0227] In some embodiments, the initial appearance information of the object under investigation and the appearance information (standard appearance information) of the object under investigation generated by the method of the embodiments of the present invention can be represented by at least one of a picture, a video, a movie, and a three-dimensional model, which can be applied to display various types of objects under investigation, and can meet different collection, display, and processing needs, and has a wide range of application and versatility.

[0228] In some examples, the impact factor information includes, but is not limited to, one of building information, types of subsystems in the building, number of subsystems in the building, types of equipment included in the subsystems in the building, and number of equipment included in the subsystems in the building.

[0229] In some embodiments, the building information includes, but is not limited to, at least one of the following: the floors of the building, the number and location of the placement (implementation) spaces of similar equipment included in the subsystems of the building (i.e., including the spatial positional relationships between the architectural space and the equipment in the building, and the spatial positional relationships between the equipment in the architectural space), the size (area) of the space, the building material, and the architectural volume information.

[0230] In some embodiments, the subsystems in the building are similar to the survey waiting system described above and include, but are not limited to, at least one of a building energy system, an HVAC central control system, an air management system, a lighting system, a security system, an architectural model system, a conference system, an entry system, and a guest system.

[0231] In some embodiments, the influence factor information is obtained from data obtained by surveying users.

[0232] For example, a survey questionnaire can be issued to users, and the survey questionnaire can include information such as the number of floors in the building, the type of subsystem, the number of subsystems, the type of equipment included in each subsystem, the number of equipment, and the floor on which the equipment is located, and by calculating statistics on the data in the survey questionnaire, information on influencing factors can be obtained.

[0233] Furthermore, for example, when communicating the needs for building inspections or the needs for building remodeling to make the building more intelligent, the user can be asked for information such as the number of floors, the type of subsystems, the number of subsystems, the type of equipment included in each subsystem, the number of equipment, and the floor on which the equipment is located, and further information on influencing factors can be obtained.

[0234] The above has described the investigation target information and influencing factor information according to an embodiment of the present invention. In step 202, the embodiment of the present invention identifies the appearance information of at least one investigation target based on the investigation target information and influencing factor information. Below, the step of identifying the appearance information of at least one investigation target will be specifically described using a specific embodiment.

[0235] In some embodiments, a trained model can be utilized to identify the appearance information of the object under investigation, i.e., a "model method" is used to identify the appearance information of the object under investigation.

[0236] 13 is a schematic diagram of a step of identifying appearance information of an investigation target according to an embodiment of the present invention. As shown in FIG. 13, step 202 of identifying appearance information of at least one investigation target based on investigation target information and influencing factor information includes step 2021 of inputting initial appearance information and influencing factor information in the investigation target information of the investigation target into a first generative model, and outputting standard appearance information of the at least one investigation target.

[0237] Here, the initial appearance information of the investigation target is one of the investigation target information of the investigation target.

[0238] This allows the "first generation model" to process the "initial appearance information" and influencing factor information in the survey object information to generate standard appearance information, eliminating the need to manually select the standard appearance information of the survey object and further improving the efficiency and accuracy of appearance information generation, thereby further improving the efficiency of building inspections, reducing labor costs, and the generated model has generalizability, allowing it to generate standard appearance information that suits the on-site situation based on new, unseen data, thereby providing greater flexibility.

[0239] Furthermore, the "first generative model" may be named differently as long as it has the function of "taking initial appearance information and influencing factor information of the object of investigation as input and outputting standard appearance information of at least one object of investigation," and embodiments of the present invention do not limit the name of the model.

[0240] In some embodiments, the first generative model is trained based on at least initial appearance information of each type of survey object in the building, survey data generated according to the needs of each survey, and standard appearance information of the survey object corresponding to the survey data. For example, the survey data includes at least information on influencing factors of the survey object, and for example, the survey data includes a survey dot plot pre-generated based on the user's needs (including the influencing factor information of the survey object). Details related to the survey dot plot may refer to related art and will not be described in detail here.

[0241] When the survey object is located in a building, the survey data may cause the on-site image of the building to be inconsistent with the initial appearance information of the survey object, and different survey data may result in different on-site images of the survey object. Therefore, in an embodiment of the present invention, a first generative model is trained using the "initial appearance information of the survey object," "survey data," and "standard appearance information of the survey object corresponding to the survey data." The trained first generative model uses the "initial appearance information of the survey object" and "survey data" as input data and can output "standard appearance information of the survey object corresponding to the survey data." The "standard appearance information of the survey object corresponding to the survey data" output by the first generative model has a higher similarity to the on-site image, which is advantageous for investigators to quickly find survey objects in buildings and further improve the efficiency of building surveys.

[0242] In some embodiments, as shown in FIG. 13, when "outputting standard appearance information of at least one research object by a first generative model," step 202 further includes step 2022 of training the first generative model.

[0243] Here, the step of training a first generative model includes obtaining training samples including initial appearance information of various types of survey objects in the building, corresponding survey data generated according to the needs of each survey, and standard appearance information of the survey objects corresponding to the survey data, where the initial appearance information of the survey objects, the survey data, and the standard appearance information of the survey objects corresponding to the survey data appear as pairs and have a one-to-one correspondence, and training an initial first generative model using the training samples. Here, the standard appearance information output from the first generative model during the training process is compared with the standard appearance information previously obtained (in the training samples), and the trained first generative model is obtained so that a loss function (e.g., L1 Loss) of the first generative model is minimized.

[0244] Since the first generative model is pre-trained, step 2022 may be performed independently or in advance; that is, step 202 does not necessarily include step 2022. Therefore, in Fig. 13, step 2022 is indicated by a dashed box as an optional step, and may be omitted in the actual process of identifying the appearance information of the object of investigation. Furthermore, Fig. 13 may further include other steps, and the present invention is not limited thereto.

[0245] In some embodiments, before training the first generative model using the training sample, feature extraction is performed on the survey data in the training sample, for example, by extracting features of the survey data using a graph neural network (GNN) and two fully connected layers (FC), and then training the first generative model based on the extracted survey data features and the initial appearance information and standard appearance information of the corresponding survey object.

[0246] Accordingly, in the process of using the trained first generative model, a graph neural network and a two-layer fully connected layer are introduced before the trained first generative model, and in the process of identifying the corresponding standard appearance information of the object to be investigated using the initial appearance information of the object to be investigated and the investigation data, the initial appearance information of the object to be investigated is input as the first input data to the first generative model, and at the same time the investigation data is input into the graph neural network. After feature extraction by the two-layer fully connected layer, the output of the second fully connected layer is used as the second input data of the first generative model, thereby outputting the corresponding standard appearance information of the object to the first generative model.

[0247] In some embodiments, the first generative model comprises a generative adversarial network (GAN) or a three-dimensional generative adversarial network (3D-GAN).

[0248] An example will be described in which the first generative model includes a generative adversarial network.

[0249] FIG. 14 is a schematic diagram illustrating generation of appearance information of a search target using a generative adversarial network (GAN) according to an embodiment of the present invention.

[0250] As shown in FIG. 14 , for a target object to be investigated, first, initial appearance information of the target object is obtained, then the initial appearance information of the target object is used as input data 1, and information on influencing factors affecting the appearance of the target object is used as input data 2, and these are input to a trained generative adversarial network (GAN), and the generative adversarial network is made to output standard appearance information of the target object.

[0251] If the first generative model includes a generative adversarial network (GAN), when training the first generative model, the "initial appearance information of the survey object" and "standard appearance information of the survey object corresponding to the survey data" in the training sample are two-dimensional data, and when using the trained first generative model, the "two-dimensional" initial appearance information of the survey object and influencing factor information are input into the first generative model, and the first generative model is caused to output the "two-dimensional" standard appearance information of the survey object.

[0252] FIG. 15 is a schematic diagram of training a first generative model including a generative adversarial network (GAN).

[0253] As shown in FIG. 15, the training process includes the following steps:

[0254] S1: The initial appearance information of each type of survey object in the training sample is used as the first input data, and each type of survey object only requires one piece of initial appearance information. The survey data for various cases is used as the second input data. The standard appearance information of the survey object, which corresponds one-to-one with the second input data, is used as the first output data, and the number of first output data corresponds one-to-one with the number of second input data.

[0255] S2: A graph neural network (GNN) and a two-layer fully connected layer (FC) are introduced before the generative adversarial network (GAN) to be trained. During training, the survey data is first input to the graph neural network, and features of the survey data are extracted through the two-layer fully connected layer, and the features of the survey data are input to the trained generative adversarial network. The initial appearance information of the survey object is also input to the trained generative adversarial network. The trained generative adversarial network generates standard appearance information (second output data) corresponding to the survey data of the survey object through computation.

[0256] S3: Select L1 Loss as the loss function.

[0257] S4: Perform analytical calculations based on the standard appearance information (second output data) output from the trained generative adversarial network and the corresponding standard appearance information (first output data) in the training samples to update the loss function, and repeat the above training process to continuously provide data so as to minimize the loss function, thereby obtaining a trained generative adversarial network.

[0258] If the first generative model includes a three-dimensional generative adversarial network, during training in the first generation mode, the "initial appearance information of the subject of investigation" and the "standard appearance information of the subject of investigation corresponding to the investigation data" in the training sample are three-dimensional model data, and when using the trained first generative model, the three-dimensional model data of the initial appearance information of the subject of investigation and influencing factor information are input into the first generative model, and the first generative model is caused to output three-dimensional model data of the standard appearance information of the subject of investigation.

[0259] FIG. 16 is a schematic diagram of training a first generative model including a 3D generative adversarial network (3D-GAN).

[0260] As shown in FIG. 16, the training process includes the following steps:

[0261] S1: The three-dimensional model data of the initial appearance information of each type of research object in the training sample is used as the first input data, and each type of research object only requires one set of three-dimensional model data of the initial appearance information, and the research data of various cases is used as the second input data, and the three-dimensional model data of the standard appearance information of the research object, which corresponds one-to-one with the second input data, is used as the first output data, and the number of the first output data corresponds one-to-one with the number of the second input data.

[0262] S2: A format conversion module is introduced before the 3D generative adversarial network (3D-GAN) to be trained, and the above first input data is input into the format conversion module to convert the first input data into a data format that can be supported by the 3D generative adversarial network.

[0263] S3: A graph neural network (GNN) and two fully connected layers (FC) are introduced before the 3D generative adversarial network to be trained. During training, the survey data is first input to the graph neural network (GNN), and features of the survey data are extracted through the two fully connected layers (FC). The survey data features are then input to the trained 3D generative adversarial network. The first input data, format-converted by the format conversion module, is also input to the trained 3D generative adversarial network. The trained 3D generative adversarial network generates 3D model data (second output data) of standard appearance information corresponding to the survey data of the survey target through computation.

[0264] S4: Select L1 Loss as the loss function.

[0265] S5: Perform analytical calculations based on the second output data output from the trained 3D generative adversarial network and the corresponding 3D model data of standard appearance information in the training sample (first output data), update the loss function, and repeat the training process to continuously provide data so as to minimize the loss function, thereby obtaining a trained 3D generative adversarial network.

[0266] Therefore, the use of a "generative adversarial network" and a "3D generative adversarial network" differs in that the dimensions of the model input data and output data are different, but the types of the input data and output data are the same, which can increase the diversity of appearance information generated using the model and is advantageous for meeting different usage needs and usage scenarios.

[0267] In some embodiments, the method further comprises obtaining initial appearance information of the study object.

[0268] In some embodiments, the step of obtaining the initial appearance information of the research object includes searching for corresponding appearance information from a database as the initial appearance information of the research object based on identification information of the research object.

[0269] For example, the identification information of all survey objects in a building and the initial appearance information corresponding to each identification information are stored in a database in advance, and when obtaining the initial appearance information of a survey object, the input identification information of the survey object is received, and matching is performed in the database based on the identification information, and if it is matched with the initial appearance information corresponding to the input identification information, the matched initial appearance information can be output to obtain the initial appearance information of the survey object.

[0270] Furthermore, an output format of the initial appearance information may be input. For example, when identification information of the object to be investigated is input, or when multiple formats of initial appearance information corresponding to the input identification information are matched, information instructing the "output format of the initial appearance information" is received, and the initial appearance information of the object to be investigated in the format corresponding to the instruction is output.

[0271] In some embodiments, the network may further be searched for corresponding appearance information based on the identification information of the search object as the initial appearance information of the search object, for example, the network may be searched for by a search engine.

[0272] In some embodiments, the step of obtaining initial appearance information of the research object includes generating initial appearance information of the research object based on identification information of the research object and a second generative model.

[0273] For example, the second generative model is, for example, a GPT model, and by inputting the identification information of the object of investigation into the second generative model as input data, the second generative model is caused to output corresponding initial appearance information.

[0274] For example, the second generative model is pre-trained using training samples. The training samples for training the second generative model include multiple sets of training data, each set of training data including, for example, predefined identification information of the research object and initial appearance information of the corresponding research object, thereby allowing the trained second generative model to output corresponding initial appearance information based on the input identification information of the research object. Alternatively, the training samples for training the second generative model include multiple sets of training data, each set of training data including, for example, predefined identification information of the research object and initial appearance information of various forms of the corresponding research object. This allows the trained second generative model to output initial appearance information of a form corresponding to the research object based on the input identification information of the research object and the output form of the initial appearance information.

[0275] In some embodiments, the identification information of the survey object includes, for example, at least one of a name and an ID, and the output form of the initial appearance information of the survey object includes, for example, at least one of a picture, a video, a video, and a three-dimensional model.

[0276] FIG. 17 is a schematic diagram of a process of searching for initial appearance information from a database according to an embodiment of the present invention.

[0277] As shown in FIG. 17, the process includes the following steps:

[0278] S1: Receive identification information of the investigation target to be investigated.

[0279] S2: The identification information received in S1 is matched with the stored data in the database, and various types of initial appearance information of the survey objects are stored in the database, and the initial appearance information of the various survey objects each has at least one form of a picture, a video, a video, or a three-dimensional model, and the initial appearance information of each survey object is stored in association with ``identification information of the survey object.''

[0280] S3: Output initial appearance information of the investigation target that matches the identification information received in S1. The output initial appearance information may be in the form of one or more of a picture, a video, a video, and a three-dimensional model.

[0281] For example, the database stores initial appearance information of survey objects such as energy consumption subsystems, lighting subsystems, and machine room subsystems, and for each survey object, the initial appearance information includes stored data in four forms: pictures, videos, images, and three-dimensional models.

[0282] This allows the matched initial appearance information to be combined based on the influencing factor information, and further allows standard appearance information of the investigation target to be obtained.

[0283] In the above embodiment, the initial appearance information of the object to be investigated is obtained by searching a database or using a second generative model, eliminating the need to manually select the initial appearance information of the object to be investigated, and improving the efficiency and accuracy of obtaining the initial appearance information, thereby improving the efficiency and accuracy of generating standard appearance information of the object to be investigated, and further improving the efficiency of the investigation.

[0284] In some embodiments, in addition to the trained model, a "mapping" may also be used to identify the appearance information of the object under investigation.

[0285] In some embodiments, "mapping" refers to stitching together the survey object information (e.g., respective photographs) for each survey object and related survey objects associated with the survey object into standard appearance information (e.g., stitched photographs) that includes the survey object and related survey objects.

[0286] 18 is a schematic diagram of another step of identifying appearance information of an investigation target according to an embodiment of the present invention. As shown in FIG. 18, step 202 of identifying appearance information of at least one investigation target based on investigation target information and influencing factor information includes step 2023 of acquiring investigation target information of the investigation target, including initial appearance information of the investigation target and each of a plurality of related investigation targets related to the investigation target, and step 2024 of generating standard appearance information including the plurality of investigation targets based on the initial appearance information and influencing factor information of the investigation target and each of a plurality of related investigation targets related to the investigation target.

[0287] In contrast to the embodiment in which standard appearance information is identified using a generative model, the “mapping” embodiment in the examples of the present application does not require training a model, reducing the cost required for training the model and the difficulty of implementation.

[0288] FIG. 19 is a schematic diagram illustrating the generation of standard appearance information using mapping according to an embodiment of the present invention.

[0289] As shown in FIG. 19, the method includes the following steps.

[0290] S1: Receive identification information and influencing factor information of the investigation target to be investigated.

[0291] S2: The identification information received in S1 is matched with the stored data in the database, and the initial appearance information of various types of survey objects is stored in the database, and the initial appearance information of each survey object is stored in association with the ``identification information of the survey object.''

[0292] S3: Output the initial appearance information of the investigation object that matches the identification information received in S1.

[0293] For example, the database stores initial appearance information of the objects of investigation, such as air contactors of the energy consumption subsystem, single-phase power meters, three-phase power meters, communication gateways of the fuse and lighting subsystems, two-way switch modules, eight-way switch modules, and sixteen-way switch modules. When identification information of the air contactors and three-phase power meters is received in S1, the initial appearance information of the air contactors and three-phase power meters can be obtained by matching them in the database.

[0294] S4: Combine the initial appearance information output in S3 with the influencing factor information received in step S1.

[0295] S5: The combined data is output as standard appearance information of the object to be investigated.

[0296] This eliminates the need to manually select standard appearance information for the object of investigation, is easy to implement, and can quickly obtain standard appearance information for the object of investigation, improving the efficiency and accuracy of generating appearance information, improving the efficiency of building investigations, and reducing labor costs.

[0297] In some embodiments, in step 2023, the initial appearance information of each of the research object and a plurality of related research objects related to the research object is obtained by querying a database, for example, by querying using the identification information of the research object and the related research objects. Specifically, reference can be made to the embodiments related to "obtaining initial appearance information of the research object," and the description will not be repeated here.

[0298] In some embodiments, the "standard appearance information of multiple investigation targets" obtained in step 2024 includes the investigation target and related investigation targets related to the investigation target. For example, taking the investigation target "switchboard" as an example, related investigation targets related to the "switchboard" include "air switchgear", so initial appearance information of the "switchboard" and initial appearance information of the "air switchgear" are respectively obtained in step 2023, and the standard appearance information generated in step 2024 simultaneously includes the two investigation targets, "switchboard" and "air switchgear".

[0299] In some embodiments, multiple related research objects related to a research object are identified at least by the relationship between the research objects. For example, according to common sense, there is a certain relationship between the research objects "switchboard" and "air switchgear." Therefore, if the research object is "switchboard," the "air switchgear" can be identified as a related research object of the "switchboard." For details regarding the relationship, please refer to the above. We will not repeat the explanation here.

[0300] In some embodiments, the information for identifying multiple related investigation objects related to the investigation object may further include influencing factor information. That is, multiple related investigation objects related to the investigation object are identified by the influencing factor information and the influencing relationship between the investigation objects. For example, based on the influencing relationship between unique investigation objects, it is determined that the investigation object "distribution board" has an influencing relationship with both an "air switch" and a "power supply circuit." At the same time, the influencing factor information "whether or not an energy consumption subsystem is included" affects the identification of the related investigation object. If the investigation object is a "distribution board" and the influencing factor information is "not included in an energy consumption subsystem," it is determined that the related investigation objects related to the "distribution board" include an "air switch" and a "power supply circuit." The finally generated standard appearance information includes a distribution board with an air switch and a power supply circuit. If the investigation object is a "distribution board" and the influencing factor information is "equipped with an energy consumption subsystem," it is identified that the related investigation objects related to the "distribution board" include an "air switch," a "power supply circuit," and a "three-phase power meter" (energy consumption subsystem), and the finally generated standard appearance information includes a distribution board with an air switch, a power supply circuit, and a three-phase power meter (energy consumption subsystem). In other words, the related investigation objects include not only unique related investigation objects (e.g., the air switch, power supply circuit, etc.) but also related investigation objects (e.g., the three-phase power meter, etc.) identified according to changes in the actual situation (as represented in the influencing factor information).

[0301] In some embodiments, step 2024 of generating standard appearance information including the multiple survey objects based on the initial appearance information of the survey object and the multiple related survey objects related to the survey object and the influencing factor information of each of them includes combining the initial appearance information of the survey object and the multiple related survey objects related to the survey object based on the influencing factor information, and treating the combined appearance information as standard appearance information including the multiple survey objects.

[0302] For example, for a "distribution board" as the survey target and "air switches" and "three-phase power meters" as related survey targets related to the distribution board, the number of distribution boards (number of equipment) in the influencing factor information affects the number of "air switches" and "three-phase power meters." Since one air switch is typically installed per distribution board, the number of "air switches" in the standard appearance information can be determined based on the number of distribution boards. Furthermore, the number of subsystems in the influencing factor information affects the number of "three-phase power meters." Since one power meter is typically installed per subsystem, the number of "three-phase power meters" in the standard appearance information can be determined based on the number of subsystems. Furthermore, the number of subsystems also affects the number of "air switches." When the number of subsystems reaches a certain level, the number of "air switches" also changes. For example, if the number of subsystems is four, one "air switch" may be installed, and if the number of subsystems is eight, two "air switches" may be installed.

[0303] Also, for example, if the investigation target includes a distribution board of a lighting system, the influencing factor information may include, for example, the spatial positional relationship between the "distribution board" in the building information and other equipment, and the type of subsystem. For example, since switch controllers are arranged in the distribution board and for lighting systems, the number of switch controllers is related to the number of lighting systems, the finally generated standard appearance information may be a combination of the corresponding number of switch controllers in the distribution board, or a combination of the distribution board and the corresponding number of switch controllers arranged in the distribution board, where the number of switch controllers is determined according to the number of lighting systems.

[0304] For example, if the survey object includes a distribution board and it is determined based on the association relationship that the related survey object includes an air switch and a three-phase power meter, the influencing factor information may include, for example, the spatial positional relationship between the "distribution board" in the building information and other equipment. For example, the air switch and the three-phase power meter are both located inside the distribution board. Therefore, the finally generated standard appearance information includes the combination of the distribution board and the air switch and the three-phase power meter located inside the distribution board. Furthermore, the positional relationship between both the air switch and the three-phase power meter within the distribution board can be further determined based on the association relationship between the survey objects. The content of the association relationship between the survey objects is not repeated here, but please refer to the above examples.

[0305] In this way, by using the influencing factor information as the basis for combining the initial appearance information of the survey object and related survey objects related to the survey object, and identifying the position and number of each survey object in the standard appearance information based on the influencing factor information, the efficiency and accuracy of generating standard appearance images can be improved, thereby further improving the efficiency of building surveys.

[0306] In some embodiments, the survey object information may further include specification information; for example, in step 2023, the survey object information of the survey object obtained includes initial appearance information and specification information of the survey object and each of multiple related survey objects related to the survey object.

[0307] Accordingly, in step 2024, standard appearance information including the plurality of investigation objects is generated based on the initial appearance information, standard information, and influencing factor information of each of the investigation object and the plurality of related investigation objects related to the investigation object.

[0308] In some embodiments, generating standard appearance information including multiple survey objects based on the initial appearance information, specification information, and influencing factor information of each of the survey objects and multiple related survey objects related to the survey object specifically includes: identifying initial appearance information of each of the survey object and multiple related survey objects related to the survey object based on specification information of each of the survey object and multiple related survey objects related to the survey object; combining the initial appearance information of each of the survey object and multiple related survey objects related to the survey object based on the influencing factor information; and treating the combined appearance information as standard appearance information including the multiple survey objects.

[0309] In this case, for each survey object, initial exterior images of different standards corresponding to the survey object are stored in the database. For example, for "distribution boards," initial exterior images of distribution boards with small specifications (e.g., length, width, and height of 600*450*350, units in mm) and initial exterior images of distribution boards with large specifications (e.g., length, width, and height of 600*1000*2050, units in mm) are stored in association with each other in the database, and an initial exterior image of a corresponding standard can be queried and selected from the database based on the standard information of the survey object. Also, for example, for "architectural space," initial exterior images of a "medium" standard and initial exterior images of a "large" standard are stored in association with each other in the database, and an initial exterior image of a corresponding standard can be queried and selected from the database based on the standard information of the survey object.

[0310] For example, assume that the specification information for a "distribution panel" being the survey target is "medium-sized," that related survey targets related to the "distribution panel" include a "switch" and a "three-phase power meter," that the specification information for the "switch" is "compact," and that the specification information for the "three-phase power meter" is "compact," and that the influencing factor information includes the spatial positional relationship between the distribution panel and the switch and the spatial positional relationship between the distribution panel and the three-phase power meter, e.g., that both the switch and the three-phase power meter are installed within the distribution panel. The initial appearance information identified based on the specification information includes the initial appearance information for the medium-sized distribution panel, the initial appearance information for the small switch, and the initial appearance information for the small three-phase power meter. Furthermore, the standard appearance information generated by combining the identified initial appearance information based on the spatial positional relationship between the equipment (influencing factor information) indicates that the small switch and the small three-phase power meter are installed within the medium-sized distribution panel. The positional relationship between the small switch and the small three-phase power meter within the medium-sized distribution panel can be identified based on the relationship between the two survey targets, the small switch and the small three-phase power meter.

[0311] This allows the standard information to be used as one of the grounds for selecting initial appearance information, the influencing factor information to be used as the grounds for combining the initial appearance information of the survey object and related survey objects, and by identifying the position and number in the standard appearance information of each survey object based on the influencing factor information, the efficiency and accuracy of generating standard appearance images can be improved, thereby further improving the efficiency of building surveys.

[0312] In some embodiments, the term "mapping" as used herein also applies to the combination of multiple quantities of a single type of research object.

[0313] For example, if it is determined based on influencing factor information that an electrical control panel contains 10 switches, the initial appearance information of the 10 switches can be combined to obtain standard appearance information for an electrical control panel having 10 switches.

[0314] Below, we will provide another specific example to explain the influence of influencing factor information on the standard exterior image of the object of investigation. Table 9 shows the influence of different influencing factor information on the generation factors of the standard exterior image of the object of investigation. In Table 9, we will explain by taking as an example that the "influencing factor information" includes building information (floor, number of spaces where the same type of equipment included in the building's subsystems is arranged, and space size), the type and number of subsystems in the building, and the type and number of equipment included in the subsystems in the building.

[0315] In Table 9, if the data in the first row indicates that "there is one equipment placement space on one floor, the size of the space is "normal", one lighting system is installed in the space, and the subsystem includes 40 lighting equipment", the generated standard appearance image will include one cabinet box, and each cabinet box will include 40 lighting equipment.

[0316] The data in the other rows indicate that when one or more of the survey data changes, the corresponding standard appearance image will also change for the data in the first row. For example, if the "space size" in the data in the second row changes to "large" and the remaining survey data remains unchanged, the generated standard appearance image may include two cabinet boxes, each with 20 lighting fixtures inside. The data in the other rows is similar, and a detailed description of each will be omitted here.

[0317] According to the above embodiment, for the object to be investigated in a building, the appearance information of the object to be investigated is generated based on the corresponding object information and influencing factor information. This makes it possible to obtain the appearance information of the object to be investigated that matches the image of the investigation site, thereby improving the accuracy of the appearance information generation. Furthermore, the present invention can automatically generate the appearance information of the object to be investigated, eliminating the need for professional engineers to manually select it, thereby improving the efficiency of appearance information generation and saving investigation costs.

[0318] Furthermore, by using appearance information that is suited to the site of investigation, it is advantageous for investigators to quickly find the object of investigation that they should investigate, improving the efficiency of building investigations and reducing investigation costs.

[0319] <Example of Second Aspect> An example of the second aspect of the present invention provides a building survey method. FIG. 20 is a block diagram of the building survey method according to an embodiment of the present invention. As shown in FIG. 20, the building survey method 300 generates a survey object. The generation method includes step 301, which may employ the building survey object generation method provided in the above-described first aspect according to an embodiment of the present invention, and step 302, of surveying the generated survey object.

[0320] Here, the building survey target generation method described in the first aspect of the embodiment of the present invention is applied to step 301 of the building survey method 300, and specifically, reference is made to the description of the embodiment of the first aspect of the present invention, and overlapping content will not be specifically explained.

[0321] In some embodiments, step 302 of investigating the generated survey objects includes transmitting the generated survey objects to a terminal equipment, receiving field data corresponding to the survey objects transmitted from the terminal equipment, obtaining standard data corresponding to each of the survey objects, and determining whether each of the survey objects meets the standard based on the field data and the standard data.

[0322] This allows researchers to quickly obtain survey subjects via terminal equipment and upload field data corresponding to each survey subject in a timely manner, which is advantageous in improving the efficiency of field data collection and further in improving survey efficiency.

[0323] Furthermore, the on-site data and the standard data include image data or three-dimensional model data.

[0324] This allows the display and processing formats of field data and standard data to be selected as needed, which is advantageous in improving the efficiency of data matching and further improves the efficiency of investigating the survey target.

[0325] In some embodiments, the building survey method 300 further includes generating appearance information of a building survey object. Figure 21 is a schematic diagram of the generating appearance information of a building survey object according to an embodiment of the present invention. As shown in Figure 21, the generating appearance information of a building survey object includes step 303 of generating appearance information of at least one survey object to be surveyed, which may employ the "method for generating appearance information of a building survey object" in the building survey object generation method provided in the first aspect of the embodiment of the present invention, and step 304 of surveying the at least one survey object based on the appearance information.

[0326] This allows the efficiency and accuracy of building inspections to be improved by investigating the target based on appearance information, thereby reducing inspection costs.

[0327] In some embodiments, the "survey object" targeted by the method for generating appearance information of a building survey object according to an embodiment of the present invention is generated based on the "building survey object generation method" in the above-described embodiment of the present invention, i.e., the steps in Figures 20 and 21 can be applied in combination. In this case, for example, first, a survey object is generated in step 301, then step 303 is executed to generate appearance information of at least one survey object to be surveyed, and then step 304 is executed to survey the at least one survey object based on the appearance information.

[0328] In some embodiments, the "survey object" targeted by the method for generating appearance information of a building survey object according to an embodiment of the present invention may be a pre-set object, a default object, or an object obtained from data obtained by surveying users, and the present application is not limited thereto. For example, in the process of surveying users, the survey object may be identified according to the user's modification needs for making the building intelligent. For example, if the user needs to modify the lighting system, the survey object may be identified to include lamps, controllers or electrical control panels corresponding to the lamps, etc.

[0329] Furthermore, step 304 of inspecting the at least one inspection object based on the appearance information includes transmitting appearance information of the at least one inspection object to a terminal equipment, receiving site data corresponding to the at least one inspection object transmitted from the terminal equipment, and determining whether the at least one inspection object meets a standard based on the appearance information and the corresponding standard data.

[0330] This allows investigators to obtain appearance information of the survey targets through terminal equipment, allowing them to quickly find the survey targets based on the appearance information and upload field data corresponding to each survey target in a timely manner, which is advantageous for improving the efficiency of field data collection and further for improving survey efficiency.

[0331] Furthermore, the on-site data, the standard data, and the appearance information include image data or three-dimensional model data.

[0332] This allows the data type, display format and processing format of the on-site data, standard data and appearance information to be selected as needed, which is advantageous in improving the efficiency of data matching and further improving the efficiency of investigating the survey object.

[0333] According to the above embodiment, the present invention can solve the problem of overlapping and redundant survey objects during the building survey process by generating survey objects according to the building survey object generation method described in the embodiment of the first aspect of the present invention, which is advantageous for improving survey efficiency and saving survey costs. Furthermore, by generating appearance information of the survey object and surveying the survey object based on the appearance information, the efficiency and accuracy of building surveys can be improved and survey costs can be further saved.

[0334] <Example of the third aspect> An example of the third aspect of the present invention provides a building inspection facility, the building inspection facility including: a memory for storing a computer program; and a processor that, when executed, implements any of the building inspection target generation methods provided in the above-mentioned first aspect of the embodiments of the present invention, and / or that, when executed, implements any of the building inspection methods provided in the above-mentioned second aspect of the embodiments of the present invention.

[0335] 22 is a block diagram of a building inspection system according to an embodiment of the present invention. As shown in FIG. 22, building inspection system 400 may include a processor 410 and a memory 420 coupled to processor 410. It should be noted that the diagram is exemplary and that other types of structures may be used in addition to or instead of the structure to implement telecommunications or other functions.

[0336] In one embodiment, the processor 410 may be configured to generate, for at least one system in a building awaiting investigation, a plurality of investigation targets corresponding to the at least one system awaiting investigation in a deduplication manner, wherein the generated plurality of investigation targets are each different, and the deduplication manner is realized based at least on the correspondence between the system awaiting investigation and the investigation targets.

[0337] In another embodiment, the processor 410 may be further configured to obtain survey target information and influencing factor information affecting the appearance of at least one survey target in the building, and to identify appearance information of the at least one survey target based on the survey target information and the influencing factor information.

[0338] In the embodiment of the present invention, for the implementation of the functions of the processor 410, reference may be made to the description of the relevant steps in the embodiment of the first aspect of the present invention, and the description will not be repeated here.

[0339] In another embodiment, the processor 410 may be arranged to employ the building survey object generation method provided in the above first aspect of an embodiment of the present invention to generate a survey object and to survey the generated survey object.

[0340] In another embodiment, the processor 410 may be further configured to employ the building survey object generation method provided in the above first aspect of the embodiment of the present invention to generate appearance information of the survey objects, and survey the at least one survey object based on the appearance information.

[0341] In the embodiment of the present invention, for the implementation of the functions of the processor 410, reference may be made to the description of the relevant steps in the embodiments of the first and second aspects of the present invention, and the description will not be repeated here.

[0342] The processor 410, sometimes referred to as a controller or operational control, may include a microprocessor or other processor and / or logic device, which receives inputs and controls the operation of each component of the building inspection equipment 400.

[0343] The memory 420 may be, for example, one or more of a buffer, a flash memory, a hard disk drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable device. In addition to storing various data, the memory 420 may also store programs that execute related information. The processor 410 may execute the programs stored in the memory 420 to achieve information storage, processing, and the like. The functions of the other components are similar to those of the conventional components, and detailed descriptions thereof will be omitted here. Each component of the building inspection equipment 400 may be realized by dedicated hardware, firmware, software, or a combination thereof, without departing from the scope of the present invention.

[0344] Furthermore, as shown in Fig. 22 , the building inspection equipment 400 may further include a communication module 430, an input / output unit 440, a display 450, and a power supply 460. Note that the building inspection device 400 does not necessarily have to include all of the components shown in Fig. 22 , and the building inspection device 400 may also include components not shown in Fig. 22 , and reference can be made to related art.

[0345] According to the above embodiment, by automatically generating multiple different survey targets for systems in a building awaiting investigation using a duplication elimination method, the problem of overlapping and redundant survey targets can be solved, which is advantageous for improving survey efficiency and reducing survey costs. By generating a survey target for at least one system awaiting investigation, it is necessary to conduct surveys on multiple systems awaiting investigation simultaneously, and in cases where multiple systems awaiting investigation have related or identical survey targets, it is possible to generate survey targets for the multiple systems awaiting investigation that do not overlap with each other. Furthermore, investigators can perform survey work on multiple systems awaiting investigation at once and do not need to repeatedly survey any of the survey targets, which is advantageous for further improving survey efficiency.

[0346] Furthermore, by generating appearance information of the object to be investigated in a building based on the corresponding object information and influencing factor information, it is possible to obtain appearance information of the object to be investigated that matches the image of the investigation site, thereby improving the accuracy of the generation of the appearance information.In addition, the present invention can automatically generate appearance information of the object to be investigated, eliminating the need for professional engineers to manually select it, improving the efficiency of generating appearance information and saving on investigation costs.Furthermore, by using appearance information that matches the investigation site, it is advantageous for investigators to quickly find the object to be investigated, improving the efficiency of building investigations and further saving on investigation costs.

[0347] Furthermore, the present invention can solve the problem of overlapping and redundant survey objects during building surveys by generating survey objects according to the building survey object generation method described in the examples of the first aspect of the present invention, which is advantageous for improving survey efficiency and saving survey costs.Furthermore, by generating appearance information of the survey object and surveying the survey object based on the appearance information, the efficiency and accuracy of building surveys can be improved and survey costs can be further saved.

[0348] An embodiment of the present invention further provides a computer-readable program, which, when executed, causes a computer to perform any of the building survey object generation methods described in the embodiment of the first aspect of the present invention and / or any of the building survey methods described in the embodiment of the second aspect of the present invention.

[0349] An embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, the computer-readable program causing a computer to execute any one of the building survey object generation methods described in the embodiment of the first aspect of the present invention and / or any one of the building survey methods described in the embodiment of the second aspect of the present invention.

[0350] An embodiment of the present invention further provides a computer program product, which, when executed by a processor, causes the computer to perform any of the building survey object generation methods described in the embodiments of the first aspect of the present invention and / or any of the building survey methods described in the embodiments of the second aspect of the present invention.

[0351] The above-described apparatus and methods according to the embodiments of the present invention may be realized in hardware or a combination of hardware and software. The present invention also relates to such computer-readable programs that, when executed by logic components, cause the logic components to realize the above-described apparatus or components, or to implement the various methods or steps described above.

[0352] The embodiment of the present invention further relates to a storage medium for storing the above program, such as a hard disk, a magnetic disk, an optical disk, a DVD, or a flash memory.

[0353] <Embodiment of the Fourth Aspect> An embodiment of the fourth aspect of the present invention provides a building inspection system including the building inspection equipment according to the third aspect of the present invention.

[0354] Furthermore, the limitations on each step of the present invention do not limit the order of the steps, provided that they do not affect the implementation of the specific solution. A step written in front may be executed first, later, or even simultaneously. As long as the solution can be implemented, any of these steps should be considered to fall within the scope of protection of the present invention.

[0355] Although the present invention has been described above with reference to specific embodiments, those skilled in the art will understand that these descriptions are merely illustrative and do not limit the scope of protection of the present invention. Those skilled in the art can make various modifications and amendments to the present invention based on the spirit and principles of the present invention, and these modifications and amendments also fall within the scope of the present invention.

Claims

1. A building survey object generation method, comprising: for at least one system in a building awaiting survey, generating a plurality of survey objects corresponding to the at least one system awaiting survey using a deduplication method, wherein the generated plurality of survey objects are different from each other, and the deduplication method is realized based at least on the correspondence between the system awaiting survey and the survey objects.

2. The method of claim 1, wherein generating a plurality of investigation targets corresponding to the at least one system to be investigated using the deduplication method includes outputting a plurality of investigation targets corresponding to the at least one system to be investigated using a pre-trained investigation target generation / deduplication model, and the investigation target generation / deduplication model is trained based at least on the correspondence between the systems to be investigated and the investigation targets and pre-set investigation target deduplication rules.

3. The method of claim 2, wherein the investigation target generation / deduplication model is trained based on correspondence relationships between systems to be investigated and investigation targets, predetermined investigation target deduplication rules, and predetermined association relationships between the investigation targets, the investigation target generation / deduplication model includes a graph neural network model, and generating multiple investigation targets corresponding to the at least one system to be investigated using the deduplication method includes outputting the multiple investigation targets and the association relationships between the multiple investigation targets using the investigation target generation / deduplication model.

4. The method of claim 2, further comprising training a survey target generation and deduplication model, wherein training the survey target generation and deduplication model comprises training an initial model using previously acquired training samples to obtain the survey target generation and deduplication model, and the training samples include at least a system to be surveyed, a survey target, and a correspondence between the system to be surveyed and the survey target.

5. The method of claim 1, wherein generating a plurality of survey targets corresponding to the at least one system to be investigated using the deduplication method includes: identifying a survey target corresponding to each system to be investigated in the at least one system to be investigated based on a pre-generated lookup table; and deduplicating all of the identified survey targets to obtain a plurality of survey targets corresponding to the at least one system to be investigated, wherein the lookup table includes at least a correspondence relationship between the systems to be investigated and the survey targets.

6. The method of claim 5, wherein the lookup table further includes association relationships between research subjects, and the method further includes identifying association relationships between the plurality of research subjects based on the lookup table.

7. The method according to claim 3 or 6, further comprising determining a search order for the plurality of search objects based on a relationship between the search objects, wherein the relationship between the search objects includes at least one of an electrical relationship, a mechanical relationship, and a spatial positional relationship between the search objects.

8. The method of claim 1, further comprising: acquiring survey target information of at least one survey target in a building and information on influencing factors that affect the appearance of the survey target; and identifying appearance information of the at least one survey target based on the survey target information and the influencing factor information.

9. The method of claim 8, wherein the investigation target information includes initial appearance information of the investigation target, and identifying appearance information of the at least one investigation target based on the investigation target information and the influence factor information includes inputting the initial appearance information of the investigation target and the influence factor information into a first generative model and outputting standard appearance information of the at least one investigation target, and the first generative model includes a generative adversarial network (GAN) or a three-dimensional generative adversarial network (3D-GAN).

10. The method of claim 9, wherein the first generative model is trained based on at least initial appearance information of each type of survey object in the building, survey data generated for each survey need, and standard appearance information of the survey object corresponding to the survey data, and the survey data includes information on influencing factors of the survey object.

11. The method of claim 10, further comprising training the first generative model, wherein training the first generative model comprises: obtaining training samples including initial appearance information of various types of survey objects in the building, corresponding survey data generated according to the needs of each survey, and standard appearance information of the survey objects corresponding to the survey data; and training an initial first generative model using the training samples; wherein the standard appearance information output from the first generative model is compared with the standard appearance information obtained in advance, and the trained first generative model is obtained so that a loss function of the first generative model is minimized.

12. The method of claim 9, wherein obtaining the initial appearance information of the object under investigation includes: searching for corresponding appearance information from a database as the initial appearance information of the object under investigation based on the identification information of the object under investigation; or generating the initial appearance information of the object under investigation based on the identification information of the object under investigation and a second generative model.

13. The method of claim 8, wherein the investigation target information includes initial appearance information of the investigation target and each of a plurality of related investigation targets related to the investigation target, and identifying the initial appearance information of the at least one investigation target based on the investigation target information and the influencing factor information includes: obtaining initial appearance information of the investigation target and each of a plurality of related investigation targets related to the investigation target; and generating standard appearance information including the plurality of investigation targets based on the initial appearance information of the investigation target and each of a plurality of related investigation targets related to the investigation target and the influencing factor information.

14. The method of claim 13, wherein generating standard appearance information including the multiple investigation targets based on the initial appearance information of the investigation target and multiple related investigation targets related to the investigation target and the influencing factor information includes: combining the initial appearance information of the investigation target and multiple related investigation targets related to the investigation target based on the influencing factor information; and treating the combined appearance information as standard appearance information including the multiple investigation targets.

15. A method according to any one of claims 8 to 14, wherein the impact factor information includes at least one of building information, types of subsystems in a building, the number of subsystems in a building, types of equipment included in the subsystems in a building, and the number of equipment included in the subsystems in a building, and the impact factor information is obtained from data obtained by surveying users.

16. The method of any one of claims 8 to 14, wherein the survey target includes at least one of architectural spaces within a building, building materials, ducts, wiring, equipment within a building, and the implementation or placement of said equipment, and / or the building information includes at least one of the number, location, spatial size, architectural materials, and architectural volume information of placement spaces for similar equipment included in a building subsystem, and / or the system to be surveyed and the building subsystem each include at least one of a building energy system, an HVAC centralized control system, an air management system, a lighting system, a security system, an architectural model system, a conference system, an entry system, and a guest system.

17. The method of claim 8, further comprising representing the plurality of survey objects and the appearance information, respectively, with at least one of a picture, a video, text, a map, and a three-dimensional model.

18. The method of claim 1, further comprising identifying the at least one system to be investigated according to a user's needs, wherein identifying the at least one system to be investigated according to the user's needs comprises: obtaining a user's needs table including a plurality of systems to be investigated and the user's markers for the systems to be investigated; and identifying the systems to be investigated marked by the user as the at least one system to be investigated, and wherein the method further comprises formatting the needs table to obtain graph data corresponding to the at least one system to be investigated.

19. The method of claim 1, further comprising: obtaining volume information of a building; and updating survey targets in the building based on the volume information.

20. A building survey method comprising: generating a survey object using the building survey object generation method according to any one of claims 1 to 19; and surveying the generated survey object.

21. The method of claim 20, wherein inspecting the generated survey objects includes: transmitting the generated multiple survey objects to a terminal equipment; receiving site data corresponding to the multiple survey objects transmitted from the terminal equipment; acquiring standard data corresponding to each of the multiple survey objects; and determining whether each of the survey objects conforms to the standard based on the site data and the standard data, wherein the site data and the standard data each include image data or three-dimensional model data.

22. The method of claim 20, wherein the building inspection method further includes generating appearance information of at least one inspection object to be inspected, and inspecting the generated inspection object includes inspecting the at least one inspection object based on the appearance information.

23. The method of claim 22, wherein inspecting the at least one survey object based on the appearance information includes: transmitting appearance information of the at least one survey object to a terminal equipment; receiving site data corresponding to the at least one survey object transmitted from the terminal equipment; and determining whether the at least one survey object conforms to a standard based on the appearance information and corresponding standard data, wherein the appearance information includes image data or three-dimensional model data.

24. A building inspection facility comprising: a memory in which a computer program is stored; and a processor that, when executing the computer program, causes the building inspection target generation method according to any one of claims 1 to 18 and / or the building inspection method according to any one of claims 19 to 23 to be realized.

25. A building inspection system comprising the building inspection equipment according to claim 24.

Images