Building decision generation method and device and electronic equipment

Abstract

The invention discloses a building decision generation method and device and electronic equipment. The method comprises the following steps: determining a target building demand, wherein a candidate demand type in a preset demand type table is the same as a demand type to which the target building demand belongs; determining at least one target building object from at least one candidate building object according to the target building requirement; determining at least one target space-time code and at least one target multi-modal data code of the target building object; and generating a target building decision meeting the target building demand on the basis of the target space-time code and the target multi-modal data code. By adopting the technical scheme of the invention, the acquisition of the data source is more diversified when the building requirements of the target building are met, and the comprehensive utilization degree of the data is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building management, and in particular to a building decision making method, device and electronic equipment. Background Art

[0002] Building management is the comprehensive management of various objects in a building, including various equipment, spaces, and systems in the building. Good building management can quickly respond to various needs in the building, generate corresponding decisions for management personnel to maintain, avoid building damage and affect the user experience in the building, but the existing building management methods are relatively single in terms of data source acquisition and low data utilization, resulting in the problem of low building management level. Summary of the invention

[0003] The present invention provides a building decision making method, device and electronic equipment to solve the problem that the existing building management method has a relatively single data source acquisition and a low degree of data utilization, resulting in a low level of building management.

[0004] According to one aspect of the present invention, a method for generating building decisions is provided, the method comprising:

[0005] Determine the target building demand, and there is a candidate demand type in the preset demand type table that is the same as the demand type to which the target building demand belongs;

[0006] According to the target building requirements, at least one target building object is determined from at least one candidate building object, where the target building object is a candidate building object involved in answering the target building requirements, and the candidate building object is a pre-matched device, system or pre-defined space range of the target building;

[0007] Determine at least one target spatiotemporal code and at least one target multimodal data code of the target building object, the candidate spatiotemporal code represents the change of the multimodal data code at different acquisition times and the connection relationship between the candidate building objects, the multimodal data is at least one data type operation data acquired by the candidate building object at the same acquisition time, the determination time of each candidate spatiotemporal code corresponding to the same candidate building object is different from each other, and the determination time of each candidate multimodal data code corresponding to the same candidate building object is different from each other;

[0008] Based on the target spatiotemporal coding and the target multimodal data coding, a target building decision that meets the target building requirements is generated.

[0009] According to another aspect of the present invention, a building decision generating device is provided, the device comprising:

[0010] A demand determination module is used to determine the target building demand, where a candidate demand type exists in the preset demand type table and is the same as the demand type to which the target building demand belongs;

[0011] A target object determination module is used to determine at least one target building object from at least one candidate building object according to the target building requirements, the target building object being a candidate building object involved in answering the target building requirements, and the candidate building object being a pre-matched device, system or pre-defined space range of the target building;

[0012] A coding determination module is used to determine at least one target spatiotemporal coding and at least one target multimodal data coding of a target building object, wherein the candidate spatiotemporal coding represents the change of the multimodal data coding at different acquisition times and the connection relationship between the candidate building objects, and the multimodal data is at least one data type of operation data collected by the candidate building object at the same acquisition time. The determination times of the candidate spatiotemporal coding for the same candidate building object are different from each other, and the determination times of the candidate multimodal data coding for the same candidate building object are different from each other;

[0013] The decision generation module generates a target building decision that meets the target building requirements based on the target spatiotemporal coding and the target multimodal data coding.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0015] at least one processor; and

[0016] a memory communicatively connected to at least one processor; wherein,

[0017] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by at least one processor so that the at least one processor can execute the building decision making method according to any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to implement the building decision-making method of any embodiment of the present invention when executed by a processor.

[0019] The technical solution of the embodiment of the present invention determines the target building demand, and a candidate demand type exists in the preset demand type table that is the same as the demand type to which the target building demand belongs, so that when the user inputs the demand, it can be ensured that the application can accurately understand the target building demand, thereby improving the accuracy of the corresponding decision generated by the target building demand; by determining at least one target building object from at least one candidate building object according to the target building demand, so that when the target building demand is met, the target building object related to the target building demand can be found, so that when the target building demand is subsequently met, the data of irrelevant candidate building objects can be reduced, thereby improving the generation efficiency of the target building decision; by determining at least one target spatiotemporal code and at least one target multimodal data code of the target building object, the target building demand is generated based on the target spatiotemporal code and the target multimodal data code, so that when the target building demand is met, all the operating data of the target building object can be comprehensively utilized, and the association relationship between the target building object and other candidate building objects can be considered, thereby ensuring the accuracy of the final generation result; through the joint use of the above steps, when the building class demand of the target building is met, the acquisition of data sources can be more diversified, and the degree of data utilization is improved.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 is a flow chart of a building decision making method provided according to Embodiment 1 of the present invention;

[0023] Figure 2 is a flow chart of another building decision making method provided according to the second embodiment of the present invention;

[0024] Figure 3 is a structural schematic diagram of a building decision making device provided according to Embodiment 3 of the present invention;

[0025] Figure 4 It is a structural schematic diagram of an electronic device for implementing the building decision making method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Embodiment 1

[0029] Figure 1 A flowchart of a building decision generation method is provided for the first embodiment of the present invention. This embodiment is applicable to the situation where the accuracy of the answer to the building requirements needs to be improved. The method can be executed by a building decision generation device. The building decision generation device can be implemented in the form of hardware and / or software. The building decision generation device can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0030] S110, determining the target building demand, and there is a candidate demand type in the preset demand type table that is the same as the demand type to which the target building demand belongs.

[0031] In the existing building management system, various data types are used separately, and various businesses are basically processed based on a single data source. This makes it difficult for the existing building management system to make a more comprehensive consideration when analyzing building needs, and it is easy to ignore some data during analysis, resulting in a large difference between the final analysis results and the actual situation. Therefore, how to communicate more types of data when answering building needs, so that the answers to building needs can be more accurate, has become the main demand.

[0032] In this application, the target building requirements can be issues related to the target building, or requirements for processing an object in the target building, such as determining the lighting conditions in the target building in the previous week, or generating temperature adjustment strategies for each space in the target building on each day of the next Monday, etc.

[0033] This application is explained using question asking as an example.

[0034] In this application, after a user asks a question and obtains the user's question content, it will be determined whether the demand type to which the question content belongs is a pre-set candidate demand type. If so, the user's question content is determined to be the target building demand. Otherwise, the user may be asked to re-propose the demand, or modify the demand, etc.

[0035] By determining the target building requirements, when the user inputs the requirements, it can be ensured that the application can accurately determine the target building requirements, thereby improving the accuracy of generating decisions that meet the target building requirements.

[0036] S120: Determine at least one target building object from at least one candidate building object according to target building requirements.

[0037] The target building object is the candidate building object involved in answering the target building requirements. The candidate building object is the pre-matched equipment, system or pre-defined space range of the target building.

[0038] Candidate building objects can be equipment, systems, and spaces within the target building, such as room area, equipment rating, and building orientation.

[0039] After obtaining the target building requirements, different target building requirements involve different candidate building objects. For example, when considering the lighting requirements in the target building, audio equipment and the like are often not involved. Therefore, after obtaining the target building requirements, it is necessary to determine the target building objects corresponding to the target building requirements. The determination method can be determined by a neural network model, or based on the semantic relationship of the target building requirements and the candidate building objects corresponding to the various candidate requirement types determined in advance. For example, if the target building requirements involve lighting, it can be determined to be related to the target building objects that can emit light.

[0040] By determining at least one target building object from at least one candidate building object according to the target building demand, it is possible to find the target building object related to the target building demand when answering the target building demand, so that when answering the target building demand subsequently, the use of data of irrelevant candidate building objects can be reduced, thereby improving the efficiency of generating target building decisions, and since the data of irrelevant candidate building objects is reduced, the accuracy of the generated results can be improved when generating decisions.

[0041] S130: Determine at least one target spatiotemporal code and at least one target multimodal data code of the target building object.

[0042] The candidate spatiotemporal codes represent the changes in multimodal data codes at different acquisition times and the connection relationship between candidate building objects. Multimodal data is operating data of at least one data type acquired by the candidate building object at the same acquisition time. The determination times of various candidate spatiotemporal codes corresponding to the same candidate building object are different, and the determination times of various candidate multimodal data codes corresponding to the same candidate building object are different.

[0043] The target spatiotemporal coding may be a candidate spatiotemporal coding of the target building object. The candidate spatiotemporal coding represents the change of multimodal data coding at different acquisition times and the connection relationship between the candidate building objects. The multimodal data is at least one data type of running data collected by the candidate building object at the same acquisition time.

[0044] The candidate building object collects at least one data type of operating data at each collection moment, including but not limited to indoor temperature, equipment power, flow rate and other data, as well as video images collected in real time by the on-site license plate entry management system or camera system of digital delivery and operation and maintenance inspection personnel, work order information entered by maintenance personnel, relevant text messages sent by staff, and historical data of some equipment and systems.

[0045] Operational data can be obtained through two channels. One is that the information that can be obtained through the online system in real time will be automatically entered into the database. The other is that some information obtained offline is basically entered manually or uploaded by file.

[0046] In order to make it possible to call all the operating data of the target building object as much as possible when calling the operating data of the target building object, at least one data type of operating data collected by each candidate building object at the same collection time will be encoded, so that the operating data of each data type can be in the same dimension, so that all the operating data corresponding to each candidate building object are included in the encoding of the same dimension, so that all the operating data of the target building object can be called in subsequent use.

[0047] In addition, since each candidate building object may have a certain correlation with other candidate building objects, and the operating data of the same candidate building object at different collection times may also have a certain correlation, if the subsequent target building needs are met, the possible correlation between each candidate building object and other candidate building objects, and the correlation between the operating data of the same candidate building object at different collection times are ignored, the decision result generated for the final target building needs will result in a large error.

[0048] Therefore, the association relationship between a candidate building object and other candidate building objects, as well as the association relationship between the running data of the same candidate building object at different collection times can be determined and encoded, thereby obtaining the candidate spatiotemporal coding of each candidate building object, so that when subsequently meeting the target building needs, the relationship between the target building object and other candidate building objects, as well as the relationship between the running data of the target building object at different collection times can be considered, thereby improving the accuracy of the decision corresponding to the final target building needs.

[0049] In an optional solution, the process of determining candidate spatiotemporal codes may include steps A1-A5:

[0050] Step A1: Based on the predetermined overall object topological relationship between each candidate building object, determine the single object topological relationship of each candidate building object, where the single object topological relationship is used to describe the connection relationship between the candidate building object and other candidate building objects directly connected to the candidate building object.

[0051] Step A2: Generate spatial coding based on the topological relationship of a single object.

[0052] Step A3: Determine and generate a time series code according to the acquisition time of different candidate building object data groups by the candidate building objects.

[0053] Step A4: Map the spatial code and the temporal code to generate a spatial vector and a temporal vector of a preset dimension.

[0054] Step A5: Fuse the spatial vector with the temporal vector to generate candidate spatiotemporal codes.

[0055] For determining the candidate spatiotemporal codes of each candidate building object at different acquisition times, two aspects need to be considered. One aspect is the association between different candidate building objects. For example, when two light bulbs are connected in series, the state of one light bulb can reflect the operating state of the other light bulb. The other aspect is the association between the operating data corresponding to the same candidate building object at different acquisition times.

[0056] For this purpose, it is necessary to determine the single object topological relationship of each candidate building object based on the pre-determined overall object topological relationship between each candidate building object, so as to obtain the association relationship between each candidate building object. After obtaining the single object topological relationship of each candidate building object, the single object topological relationship can be encoded to generate the spatial encoding of each candidate building object in space, wherein the single object topological relationship is used to represent the connection relationship between each candidate building object and each adjacent candidate building object in actual situation.

[0057] In addition, the candidate building object needs to determine the collection time of different candidate building object data groups to generate a time sequence code to determine the change of the candidate building object data group of the same candidate building object at adjacent collection times. The determination of the spatial code and the time sequence code are determined in units of the collection time, that is, each collection time has a corresponding spatial code and time sequence code.

[0058] In order to enable the fusion of spatial coding and temporal coding, the spatial coding and the temporal coding are mapped to generate a spatial vector and a temporal vector of a preset dimension, and the spatial vector and the temporal vector are fused to generate a candidate space-time coding.

[0059] Based on the pre-determined overall object topological relationship between each candidate building object, the single object topological relationship of each candidate building object is determined, and the single object topological relationship is used to describe the connection relationship between the candidate building object and other candidate building objects directly connected to the candidate building object; based on the single object topological relationship, a spatial code is generated; according to the acquisition time of different candidate building object data groups by the candidate building object, a time series code is generated; the spatial code and the time series code are mapped to generate a spatial vector and a time series vector of a preset dimension, and finally the spatial vector and the time series vector are merged to generate a candidate spatiotemporal code, thereby clarifying the determination process of the candidate spatiotemporal code of each candidate building object at different acquisition times.

[0060] Optionally, the process of determining at least one candidate building object data group corresponding to each candidate building object includes:

[0061] Collect data of candidate building objects at each collection time to obtain initial candidate building object data corresponding to each collection time;

[0062] The initial candidate building object data is cleaned to obtain at least one candidate building object data group corresponding to each candidate building object.

[0063] In order to ensure the accuracy of at least one candidate building object data group corresponding to each candidate building object, it is necessary to collect data for the candidate building objects at each collection moment. After obtaining the initial candidate building object data corresponding to each collection moment, the initial candidate building object data is cleaned to remove abnormal values, dead values, blank values, etc. that may exist in the initial candidate building object data.

[0064] S140, based on the target spatiotemporal coding and the target multimodal data coding, generate a target building decision that meets the target building requirements.

[0065] After obtaining the target spatiotemporal coding and the target multimodal data coding, it is equivalent to determining the association relationship between the target building object and other candidate building objects and the time relationship between different acquisition times of the target building object, and obtaining all the operating data corresponding to the target building object. This makes it possible to comprehensively utilize all the operating data of the target building object while considering the association relationship between the target building object and other candidate building objects when meeting the needs of the target building, thereby ensuring the accuracy of the final generated results.

[0066] In an optional solution, answering the target building requirements based on the target spatiotemporal coding and the target multimodal data coding may include steps B1-B4:

[0067] Step B1. Generate at least one first feature vector and at least one second feature vector based on at least one target spatiotemporal code and at least one target multimodal data code. The first feature vector is obtained by mapping the multimodal data code in the target building object data within the same acquisition moment into a vector with the same dimension as the target spatiotemporal code, and then concatenating the vector with the target spatiotemporal code. The second feature vector is obtained by mapping the target spatiotemporal code within the same acquisition moment into a vector with the same dimension as the multimodal data code in the target building object data, and then concatenating the vector with the multimodal data code.

[0068] Step B2: input each first feature vector into the first model for processing to obtain a first result. The first model is a neural network model with prediction and classification functions.

[0069] Step B3: input each second feature vector into the second model for processing to obtain a second result, where the second model is a large language model.

[0070] Step B4: Determine the target answer based on the first result and the second result.

[0071] In order to ensure the accuracy of decision-making corresponding to demand, this application will consider two data analysis methods when meeting the demand. One is that the requirements for quantitative results are high, and the silicon-based perspective is better at such calculations, such as various predictions of various classification needs. The other is biased towards qualitative analysis, which is better at such analysis from a human perspective.

[0072] For the first type, the multimodal data encoding in the target building object data at the same acquisition moment is mapped into a vector with the same dimension as the target spatiotemporal encoding, and then concatenated with the target spatiotemporal encoding to generate the first feature vector, which is then input into a neural network model with prediction and classification functions, so as to give an answer with the help of the neural network model.

[0073] For the second type, the target spatiotemporal coding within the same acquisition moment is mapped into a vector with the same dimension as the multimodal data coding in the target building object data, and then concatenated with the multimodal data coding to generate a second feature vector, which is input into a large language model to answer the needs.

[0074] Afterwards, the first result and the second result are sorted to generate the final target answer.

[0075] At least one first eigenvector and at least one second eigenvector are generated according to at least one target spatiotemporal code and at least one target multimodal data code, each first eigenvector is input into a first model for processing to obtain a first result, each second eigenvector is input into a second model for processing to obtain a second result, and a target answer is determined based on the first result and the second result, thereby improving the accuracy of the answer result corresponding to the target building demand.

[0076] Optionally, the overall object topological relationship is generated through actual connection relationships among candidate building objects in the target building.

[0077] In an optional solution, the process of constructing the second model may include steps C1-C3:

[0078] Step C1, obtaining a first training sample set, the first training sample set including at least one first sample object, each first sample object including at least one sample candidate spatiotemporal code and at least one sample multimodal data code, the sample multimodal data code is obtained by tokenizing a sample object data group collected by the sample object at each collection moment, each sample object data group including at least one sample object data, and the data types of each sample object data in the same sample object data group are different.

[0079] Step C2: Based on the prompt word engineering, sample requirements and sample answers are generated according to the first training sample set.

[0080] Step C3: text-align the second model using sample requirements and sample answers.

[0081] Since the second model is a large language model, in order to improve the understanding ability of the large language model, the second model will be semantically trained through a batch of training sample sets obtained in advance, so that the second model can quickly understand the logical relationship between the input encoding and semantics. For this purpose, the first training sample set will be obtained, and based on the prompt word engineering, the first training sample set will be used as the basis for generating sample requirements and sample answers, and finally the second model will be text aligned through the sample requirements and sample answers. Among them, the sample requirements and sample answers are in a corresponding relationship, and both are architectural requirements.

[0082] By obtaining the first training sample set, and based on the prompt word engineering, generating sample requirements and sample answers according to the first training sample set, and finally aligning the second model with the sample requirements and sample answers, so that the second model can quickly establish the association relationship between the code and the text, thereby achieving the accuracy of the second result output by the second model.

[0083] According to the technical solution of the embodiment of the present invention, by determining the target building demand, there is a candidate demand type in the preset demand type table that is the same as the demand type to which the target building demand belongs, so that when the user inputs the demand, it can be ensured that the application can accurately understand the target building demand, thereby improving the accuracy of the corresponding decision generated by the target building demand; by determining at least one target building object from at least one candidate building object according to the target building demand, when the target building demand is met, the target building object related to the target building demand can be found, so that when the target building demand is subsequently met, the data of irrelevant candidate building objects can be reduced, thereby improving the generation efficiency of the target building decision; by determining at least one target spatiotemporal code and at least one target multimodal data code of the target building object, the target building demand is determined based on the target spatiotemporal code and the target multimodal data code, so that when the target building demand is met, all the operating data of the target building object can be comprehensively utilized, and the association relationship between the target building object and other candidate building objects can be considered, thereby ensuring the accuracy of the final generation result; through the joint use of the above steps, when the building class demand of the target building is met, the acquisition of data sources can be more diversified, and the degree of data utilization is improved.

[0084] Embodiment 2

[0085] Figure 2The present invention provides a flowchart of another method for generating building decisions. Based on the above embodiments, this embodiment further optimizes the process of determining at least one target spatio-temporal code and at least one target multi-modal data code of the target building object in the foregoing embodiments. This embodiment can be combined with each alternative in the above one or more embodiments. As Figure 2 shown, the method for generating building decisions in this embodiment may include the following steps:

[0086] S210. Determine the target building requirements, and there is a candidate requirement type in the preset requirement type table that is the same as the requirement type to which the target building requirements belong.

[0087] S220. Determine at least one target building object from at least one candidate building object according to the target building requirements.

[0088] The target building object is the candidate building object involved when answering the target building requirements, and the candidate building object is the equipment, system or pre-defined space range pre-matched by the target building.

[0089] S230. Determine at least one target spatio-temporal code and at least one target multi-modal data code corresponding to each target building object from the preset database.

[0090] The preset database contains at least one candidate building object, each candidate building object contains at least one candidate spatio-temporal code and at least one multi-modal data code, the multi-modal data code is obtained by tokenizing the candidate building object data group collected by the candidate building object at each acquisition moment, each candidate building object data group contains at least one candidate building object data, and the data types of the candidate building object data in the same candidate building object data group are different.

[0091] Tokenization is to split a continuous text into independent and meaningful units, and these units are called Tokens. These Tokens can be words, characters, sub-words or other text segments defined according to specific requirements. For example, in English text, words are usually the most common Tokens. For example, the sentence "I love apple" can be split into the Tokens "I", "love", and "apple". In some Asian languages such as Chinese, since there is no natural delimiter between Chinese characters, characters or words may be used as Tokens. For example, "我爱吃苹果" can be split into individual Chinese characters "我", "爱", "吃", "苹", and "果" according to different tokenization strategies, or split into words "我", "爱", "吃", and "苹果".

[0092] In order to improve the acquisition efficiency of at least one target spatiotemporal code and at least one target multimodal data code corresponding to each target building object, the present application stores the target building object as a storage unit in a preset database, that is, the data acquired from different channels will fall under each object. At this time, in the preset database, each candidate building object contains at least one candidate spatiotemporal code and at least one multimodal data code, and the multimodal data code is obtained by tokenizing and encoding the candidate building object data group collected by the candidate building object at each collection time.

[0093] Such a storage method makes it possible to directly obtain at least one target spatiotemporal code and at least one target multimodal data code corresponding to the target building object after determining the target building object, thereby improving the acquisition efficiency of the target spatiotemporal code and the target multimodal data code.

[0094] S240, based on the target spatiotemporal coding and the target multimodal data coding, generate a target building decision that meets the target building requirements.

[0095] According to the technical solution of an embodiment of the present invention, by determining at least one target spatiotemporal code and at least one target multimodal data code corresponding to each target building object from a preset database, after the target building object is determined, at least one target spatiotemporal code and at least one target multimodal data code corresponding to the target building object can be directly obtained, thereby improving the acquisition efficiency of the target spatiotemporal code and the target multimodal data code.

[0096] Embodiment 3

[0097] Figure 3 The present invention provides a structural block diagram of a building decision making device. This embodiment is applicable to situations where it is necessary to answer building requirements and improve the accuracy of the answer. The building decision making device can be implemented in the form of hardware and / or software. The building decision making device can be configured in an electronic device with data processing capabilities. Figure 3 As shown, the building decision making device of this embodiment may include: a demand determination module 310, a target object determination module 320, a coding determination module 330 and a decision making module 340. Among them:

[0098] A demand determination module 310 is used to determine the target building demand, where there is a candidate demand type in the preset demand type table that is the same as the demand type to which the target building demand belongs;

[0099] A target object determination module 320, configured to determine at least one target building object from at least one candidate building object according to the target building requirement;

[0100] A coding determination module 330, configured to determine at least one target spatiotemporal coding and at least one target multimodal data coding of the target building object;

[0101] The decision generation module 340 is used to generate a target building decision that meets the target building requirements based on the target spatiotemporal coding and the target multimodal data coding.

[0102] Based on the above embodiment, optionally, the decision making module 340 includes:

[0103] Generate at least one first feature vector and at least one second feature vector according to at least one target spatiotemporal code and at least one target multimodal data code;

[0104] Inputting each of the first feature vectors into a first model for processing to obtain a first result, wherein the first model is a neural network model with prediction and classification functions;

[0105] Inputting each of the second feature vectors into a second model for processing to obtain a second result, wherein the second model is a large language model;

[0106] A target building decision that meets the target building requirements is generated according to the first result and the second result.

[0107] Based on the above embodiment, optionally, the first model is a neural network model with prediction and classification functions; and the second model is a large language model.

[0108] Based on the above embodiment, optionally, the encoding determination module 330 includes:

[0109] From a preset database, at least one target spatiotemporal code and at least one target multimodal data code corresponding to the target building object are determined, the preset database contains at least one candidate building object, each candidate building object contains at least one candidate spatiotemporal code and at least one candidate multimodal data code, the multimodal data code is obtained by tokenizing a candidate building object data group collected at each collection time of the candidate building object, each candidate building object data group contains at least one candidate building object data, and the data types of each candidate building object data in the same candidate building object data group are different.

[0110] Based on the above embodiment, optionally, the process of determining the candidate space-time coding includes:

[0111] Based on the predetermined overall object topological relationship between the candidate building objects, determining a single object topological relationship of each candidate building object, wherein the single object topological relationship is used to describe a connection relationship between the candidate building object and other candidate building objects directly connected to the candidate building object;

[0112] Based on the single object topological relationship, generating a spatial code;

[0113] Generating a time series code according to the acquisition time of different candidate building object data groups by the candidate building object;

[0114] Mapping the spatial code and the temporal code to generate a spatial vector and a temporal vector of a preset dimension;

[0115] The spatial vector and the temporal vector are fused to generate candidate spatiotemporal codes.

[0116] Based on the above embodiment, optionally, the overall object topological relationship is generated by the actual connection relationship of each candidate building object in the target building.

[0117] Based on the above embodiment, optionally, the process of constructing the second model includes:

[0118] Acquire a first training sample set; generate sample requirements and sample decisions based on the first training sample set based on the prompt word engineering;

[0119] The second model is text-aligned through sample requirements and sample decisions.

[0120] Based on the above embodiment, optionally, the process of determining at least one candidate building object data group corresponding to each candidate building object includes:

[0121] Collect data of candidate building objects at each collection time to obtain initial candidate building object data corresponding to each collection time;

[0122] The initial candidate building object data is cleaned to obtain at least one candidate building object data group corresponding to each candidate building object.

[0123] The building decision generation device provided in the embodiment of the present invention can execute the building decision generation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0124] Embodiment 4

[0125] Figure 4A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0126] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0127] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0128] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a building decision generation method.

[0129] In some embodiments, the building decision generation method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the building decision generation method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the building decision generation method in any other appropriate manner (e.g., by means of firmware).

[0130] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0131] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0132] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0133] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0134] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0135] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0136] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0137] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating building decisions, characterized in that: include: Determine the target building demand, and there is a candidate demand type in the preset demand type table that is the same as the demand type to which the target building demand belongs; According to the target building requirement, determining at least one target building object from at least one candidate building object; Determining at least one target spatiotemporal code and at least one target multimodal data code of the target building object; Based on the target spatiotemporal coding and the target multimodal data coding, a target building decision that meets the target building requirements is generated.

2. The method according to claim 1, characterized in that Based on the target spatiotemporal coding and the target multimodal data coding, a target building decision that meets the target building requirements is generated, including: Generate at least one first feature vector and at least one second feature vector according to at least one target spatiotemporal code and at least one target multimodal data code; Inputting each of the first feature vectors into a first model for processing to obtain a first result; Inputting each of the second feature vectors into a second model for processing to obtain a second result; A target building decision that meets the target building requirements is generated according to the first result and the second result.

3. The method according to claim 2, characterized in that The first model is a neural network model with prediction and classification functions; the second model is a large language model.

4. The method according to claim 1, characterized in that: Determining at least one target spatiotemporal code and at least one target multimodal data code of the target building object comprises: From a preset database, at least one target spatiotemporal code and at least one target multimodal data code corresponding to the target building object are determined, the preset database contains at least one candidate building object, each candidate building object contains at least one candidate spatiotemporal code and at least one candidate multimodal data code, the multimodal data code is obtained by tokenizing a candidate building object data group collected at each collection time of the candidate building object, each candidate building object data group contains at least one candidate building object data, and the data types of each candidate building object data in the same candidate building object data group are different.

5. The method according to claim 4, characterized in that The process of determining candidate space-time codes includes: Based on the predetermined overall object topological relationship between the candidate building objects, determining a single object topological relationship of each candidate building object, wherein the single object topological relationship is used to describe a connection relationship between the candidate building object and other candidate building objects directly connected to the candidate building object; Based on the single object topological relationship, generating a spatial code; Generating a time series code according to the acquisition time of different candidate building object data groups by the candidate building object; Mapping the spatial code and the temporal code to generate a spatial vector and a temporal vector of a preset dimension; The spatial vector and the temporal vector are fused to generate candidate spatiotemporal codes.

6. The method according to claim 5, characterized in that The overall object topological relationship is generated through the actual connection relationship of each candidate building object in the target building.

7. The method according to claim 2, characterized in that The construction process of the second model includes: Obtaining a first training sample set; Based on the prompt word engineering, generating sample requirements and sample answers according to the first training sample set; The second model is text-aligned using sample requirements and sample answers.

8. The method according to claim 4, characterized in that The process of determining at least one candidate building object data group corresponding to each candidate building object includes: Collect data of candidate building objects at each collection time to obtain initial candidate building object data corresponding to each collection time; The initial candidate building object data is cleaned to obtain at least one candidate building object data group corresponding to each candidate building object.

9. A building decision making device, characterized in that: include: A demand determination module is used to determine the target building demand, where a candidate demand type exists in the preset demand type table and is the same as the demand type to which the target building demand belongs; A target object determination module, used to determine at least one target building object from at least one candidate building object according to the target building requirement; A coding determination module, used to determine at least one target spatiotemporal coding and at least one target multimodal data coding of the target building object; The decision generation module generates a target building decision that meets the target building requirements based on the target spatiotemporal coding and the target multimodal data coding.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the building decision making method according to any one of claims 1 to 8.

Images