A monitoring method and system based on building informatization

Abstract

The present invention provides a monitoring method and system based on building informatization, which relates to the field of data processing technology. In the present invention, a key information mining unit is used to perform key information mining processing on target building information model data to output a key information mining result representation vector; a key information extraction unit is used to perform at least one key information extraction processing on the key information mining result representation vector to output at least one selected key information mining result representation vector; a building quality prediction unit is used to perform building quality prediction processing on the key information mining result representation vector and each selected key information mining result representation vector to output at least two building quality prediction results; at least two building quality prediction results are fused to output a target building quality prediction result. Based on the above method, the reliability of building quality monitoring can be improved to a certain extent.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a monitoring method and system based on building informatization. Background Art

[0002] Building Information Modeling (BIM, Building Information Modeling, is a digital tool used in engineering design, construction, and management. It integrates the digital and information models of buildings, shares and transmits them throughout the life cycle of project planning, operation, and maintenance, enables engineering and technical personnel to correctly understand and efficiently respond to various building information, and provides a basis for collaborative work for design teams and various construction entities including construction and operation units. It plays an important role in improving production efficiency, saving costs, and shortening construction periods.) As a type of data in building informatization, BIM has many applications. For example, in some applications, it is necessary to analyze the quality of the corresponding building based on the data of the building information model in order to monitor the building quality. However, in the existing technology, there is a problem of low reliability in building quality monitoring. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a monitoring method and system based on building informatization, so as to improve the reliability of building quality monitoring to a certain extent.

[0004] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0005] A monitoring method based on building informatization, comprising:

[0006] Acquire request information for obtaining building information data to be processed, and parse the request information for obtaining building information data to be processed to obtain building information model description data to be matched carried in the request information for obtaining building information data to be processed, and determine a target building information model from multiple candidate building information models based on the building information model description data to be matched, and use the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed;

[0007] Using the key information mining unit included in the target building quality analysis neural network, the target building information model data is subjected to key information mining processing to output a key information mining result representation vector corresponding to the target building information model data;

[0008] Using the key information extraction unit included in the target building quality analysis neural network, perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector;

[0009] Utilizing the building quality prediction unit included in the target building quality analysis neural network, respectively performing building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors, so as to output at least two building quality prediction results corresponding to the target building information model data;

[0010] The at least two building quality prediction results are fused to output a target building quality prediction result corresponding to the target building information model, wherein the target building quality prediction result is used to reflect the building quality of the target building corresponding to the target building information model.

[0011] In some preferred embodiments, in the above-mentioned monitoring method based on building informatization, the step of using the key information mining unit included in the target building quality analysis neural network to perform key information mining processing on the target building information model data to output the key information mining result representation vector corresponding to the target building information model data includes:

[0012] Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing to form a model data feature mapping vector corresponding to the target building information model data;

[0013] The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

[0014] In some preferred embodiments, in the above-mentioned monitoring method based on building informatization, the step of using the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector includes:

[0015] Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector to output a selected key information mining result representation vector corresponding to the first key information convolution subunit, wherein the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit;

[0016] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit;

[0017] Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, at least one corresponding selected key information mining result representation vector is obtained.

[0018] In some preferred embodiments, in the above-mentioned monitoring method based on building informatization, for each other key information convolution subunit other than the first key information convolution subunit, the other key information convolution subunit is used to perform a convolution operation on the output data of the previous key information convolution subunit to output a characterization vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and the characterization vector of the key information mining result to be processed and the output data of the previous key information convolution subunit are fused to output the step of the selected key information mining result characterization vector corresponding to the other key information convolution subunit, including:

[0019] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit;

[0020] The representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit and the output data of the previous key information convolution subunit are concatenated to output the representation vector of the selected key information mining result corresponding to the other key information convolution subunit.

[0021] In some preferred embodiments, in the above-mentioned monitoring method based on building informatization, for each other key information convolution subunit other than the first key information convolution subunit, using the other key information convolution subunit, performing a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, the step includes:

[0022] For each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is filtered using the filter matrix included in the other key information convolution subunit to output a filtered key information mining result representation vector corresponding to the other key information convolution subunit;

[0023] Performing vector segmentation processing on the filtering key information mining result representation vector to form a local filtering key information mining result representation vector cluster corresponding to the filtering key information mining result representation vector, wherein the local filtering key information mining result representation vector cluster includes at least two local filtering key information mining result representation vectors;

[0024] For each of the local filtering key information mining result representation vectors, mark the local filtering key information mining result representation vector as a target local filtering key information mining result representation vector, and mark other local filtering key information mining result representation vectors other than the local filtering key information mining result representation vector as reference local filtering key information mining result representation vectors, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the target local filtering key information mining result representation vector, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector, and respectively perform association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector, and superimpose each of the association representation vectors to output the target association representation vector corresponding to the target local filtering key information mining result representation vector;

[0025] The target association representation vector corresponding to each of the local filtering key information mining result representation vectors is merged and processed with the key information mining result representation vector to be processed corresponding to the other key information convolution sub-units.

[0026] In some preferred embodiments, in the above-mentioned monitoring method based on building informatization, the step of performing correlation analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding correlation representation vector includes:

[0027] The query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector are multiplied to output a corresponding association representation vector.

[0028] In some preferred embodiments, in the above-mentioned monitoring method based on building informationization, the steps of obtaining request information for obtaining building information data to be processed, parsing the request information for obtaining building information data to be processed, so as to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed, and determining a target building information model from multiple candidate building information models based on the building information model description data to be matched, and using the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed, include:

[0029] Acquire the request information for obtaining the building information data to be processed, and parse the request information for obtaining the building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining the building information data to be processed;

[0030] Acquire model identification data corresponding to each candidate building information model among a plurality of candidate building information models, wherein the model identification data is formed in advance by performing data analysis processing on the corresponding candidate building information model to identify the corresponding candidate building information model;

[0031] Performing data matching calculations on the to-be-matched building information model description data and the model identification data corresponding to each of the candidate building information models, so as to output data matching parameters between the to-be-matched building information model description data and each of the candidate building information models;

[0032] According to the data matching parameters between the building information model description data to be matched and each of the candidate building information models, a target building information model is determined from the multiple candidate building information models, and the target building information model data corresponding to the target building information model is used as the data collection result corresponding to the request information for obtaining the building information data to be processed.

[0033] The embodiment of the present invention further provides a monitoring system based on building informatization, comprising:

[0034] A data acquisition module is used to obtain the request information for obtaining the building information data to be processed, and parse the request information for obtaining the building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining the building information data to be processed, and, based on the building information model description data to be matched, determine a target building information model from multiple candidate building information models, and use the target building information model data corresponding to the target building information model as the data acquisition result corresponding to the request information for obtaining the building information data to be processed;

[0035] A key information mining module is used to perform key information mining processing on the target building information model data using the key information mining unit included in the target building quality analysis neural network to output a key information mining result representation vector corresponding to the target building information model data;

[0036] A key information extraction module, used to use the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector;

[0037] A building quality prediction module, for using the building quality prediction unit included in the target building quality analysis neural network to respectively perform building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors, so as to output at least two building quality prediction results corresponding to the target building information model data;

[0038] The prediction result fusion module is used to fuse the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model, and the target building quality prediction result is used to reflect the building quality status of the target building corresponding to the target building information model.

[0039] In some preferred embodiments, in the above-mentioned monitoring system based on building informatization, the key information mining module is specifically used for:

[0040] Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing to form a model data feature mapping vector corresponding to the target building information model data;

[0041] The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

[0042] In some preferred embodiments, in the above-mentioned monitoring system based on building informatization, the key information extraction module is specifically used for:

[0043] Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector to output a selected key information mining result representation vector corresponding to the first key information convolution subunit, wherein the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit;

[0044] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit;

[0045] Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, at least one corresponding selected key information mining result representation vector is obtained.

[0046] A monitoring method and system based on building informationization provided by an embodiment of the present invention can utilize a key information mining unit to perform key information mining processing on target building information model data to output a key information mining result representation vector; utilize a key information extraction unit to perform at least one key information extraction processing on the key information mining result representation vector to output at least one selected key information mining result representation vector; utilize a building quality prediction unit to perform building quality prediction processing on the key information mining result representation vector and each selected key information mining result representation vector to output at least two building quality prediction results; and perform fusion processing on at least two building quality prediction results to output a target building quality prediction result. Based on the aforementioned steps, since the key information mining result representation vector and each selected key information mining result representation vector are subjected to building quality prediction processing to output at least two building quality prediction results, and then fusion processing is performed to output the target building quality prediction result, the reliability of the target building quality prediction result is higher, thereby improving the reliability of building quality monitoring to a certain extent.

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A structural block diagram of a monitoring platform based on building informatization provided in an embodiment of the present invention.

[0049] Figure 2 A flowchart of the steps of the monitoring method based on building informatization provided in an embodiment of the present invention.

[0050] Figure 3 A schematic diagram of various modules included in a monitoring system based on building informatization provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0052] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] like Figure 1 As shown, an embodiment of the present invention provides a monitoring platform based on building informatization, wherein the monitoring platform based on building informatization may include a memory and a processor.

[0054] In detail, the memory and the processor are electrically connected directly or indirectly to achieve data transmission or interaction. For example, they can be electrically connected to each other through one or more communication buses or signal lines. The memory can store at least one software function module (computer program) that can exist in the form of software or firmware. The processor can be used to execute the executable computer program stored in the memory, thereby realizing the monitoring method based on building informatization provided by the embodiment of the present invention.

[0055] For example, in some alternative implementations, the memory can be, but is not limited to, random access memory (RAM), read only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0056] For example, in some alternative implementations, the processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0057] For example, in some alternative implementations, the monitoring platform based on building informationization may be a server with data processing capabilities.

[0058] Combination Figure 2 The embodiment of the present invention also provides a monitoring method based on building informatization, which can be applied to the monitoring platform based on building informatization. The method steps defined in the process related to the monitoring method based on building informatization can be implemented by the monitoring platform based on building informatization. Figure 2 The specific process shown is explained in detail.

[0059] Step S100, obtaining request information for obtaining building information data to be processed, and parsing the request information for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed, and, based on the building information model description data to be matched, determining a target building information model from multiple candidate building information models, and using the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed.

[0060] In an embodiment of the present invention, the monitoring platform based on building informationization can obtain request information for obtaining building information data to be processed, and parse the request information for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed, and, based on the building information model description data to be matched, determine a target building information model from multiple candidate building information models, and use the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed.

[0061] Step S200, using the key information mining unit included in the target building quality analysis neural network, performs key information mining processing on the target building information model data to output a key information mining result representation vector corresponding to the target building information model data.

[0062] In an embodiment of the present invention, the monitoring platform based on building informationization can utilize the key information mining unit included in the target building quality analysis neural network to perform key information mining processing on the target building information model data to output the key information mining result representation vector corresponding to the target building information model data.

[0063] Step S300, utilizing the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector.

[0064] In an embodiment of the present invention, the monitoring platform based on building informationization can utilize the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector.

[0065] Step S400, using the building quality prediction unit included in the target building quality analysis neural network, respectively performs building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors to output at least two building quality prediction results corresponding to the target building information model data.

[0066] In an embodiment of the present invention, the monitoring platform based on building informationization can utilize the building quality prediction unit included in the target building quality analysis neural network to perform building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors (exemplarily, the building quality prediction unit may include a softmax function to map and output corresponding building quality prediction results) to output at least two building quality prediction results corresponding to the target building information model data.

[0067] Step S500: fusing the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model.

[0068] In an embodiment of the present invention, the monitoring platform based on building informationization can perform fusion processing (such as taking an average, etc.) on the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model. The target building quality prediction result is used to reflect the building quality of the target building corresponding to the target building information model.

[0069] Based on the aforementioned steps, the key information mining result representation vector and each selected key information mining result representation vector will be processed for building quality prediction to output at least two building quality prediction results, and then fused to output the target building quality prediction result, so that the reliability of the target building quality prediction result is higher, thereby improving the reliability of building quality monitoring to a certain extent, and thus improving the problem of poor reliability existing in the prior art.

[0070] For example, in some alternative implementations, the step of obtaining request information for obtaining building information data to be processed, parsing the request information for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed, and determining a target building information model from multiple candidate building information models based on the building information model description data to be matched, and using the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed, may further include the following contents, such as the specific contents included in step S110, step S120, step S130 and step S140.

[0071] Step S110: obtaining request information for obtaining building information data to be processed, and parsing the request information for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed.

[0072] In an embodiment of the present invention, the monitoring platform based on building informationization can obtain request information for obtaining building information data to be processed, and parse and process the request information for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining building information data to be processed.

[0073] Step S120: acquiring model identification data corresponding to each candidate building information model among the plurality of candidate building information models.

[0074] In an embodiment of the present invention, the monitoring platform based on building informationization can obtain model identification data corresponding to each candidate building information model among multiple candidate building information models. The model identification data is formed in advance by performing data analysis processing on the corresponding candidate building information model to identify the corresponding candidate building information model.

[0075] Step S130, performing data matching calculation on the to-be-matched building information model description data and the model identification data corresponding to each of the candidate building information models, so as to output data matching parameters between the to-be-matched building information model description data and each of the candidate building information models.

[0076] In an embodiment of the present invention, the monitoring platform based on building informatization can perform data matching calculations on the to-be-matched building information model description data and the model identification data corresponding to each of the candidate building information models, respectively, to output data matching parameters between the to-be-matched building information model description data and each of the candidate building information models, respectively (exemplarily, when the data forms of the to-be-matched building information model description data and the model identification data are the same, data similarity calculation processing can be directly performed; when both are text data, text similarity calculation can be directly performed, which may refer to text semantic similarity, etc.; when the data forms of the to-be-matched building information model description data and the model identification data are different, feature space mapping processing can be performed respectively first to obtain corresponding feature mapping description vectors, and then the vector similarity between the feature mapping description vectors can be calculated).

[0077] Step S140, determining a target building information model from the multiple candidate building information models based on the data matching parameters between the building information model description data to be matched and each of the candidate building information models, and using the target building information model data corresponding to the target building information model as the data acquisition result corresponding to the building information data to be processed acquisition request information.

[0078] In an embodiment of the present invention, the monitoring platform based on building informationization can determine a target building information model from the multiple candidate building information models based on the data matching parameters between the building information model description data to be matched and each of the candidate building information models (exemplarily, a candidate building information model with the largest corresponding data matching coefficient can be marked as the target building information model), and use the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining the building information data to be processed (i.e., achieving the purpose of collecting the target building information model from multiple candidate building information models).

[0079] Based on the above method, since the model identification data is formed in advance by performing data analysis and processing on the corresponding candidate building information models, in the process of model selection, that is, model data collection, only data matching calculations need to be performed on the model identification data. Compared with the conventional scheme that requires data matching calculations to be performed on the entire building information model description data, the amount of data for data matching calculations is greatly reduced, so that the efficiency of data matching calculations is improved, thereby improving the collection efficiency of target building information model data to a certain extent and improving the deficiencies in the prior art (especially when the building information model is relatively complex, the amount of data for the entire building information model description data will be very large, and therefore, the amount of data for data matching calculations can be reduced to a large extent).

[0080] For example, in some alternative implementations, based on the above steps, the monitoring method based on building informatization may further include the following steps:

[0081] Performing network optimization to form an optimized building information model analysis neural network;

[0082] Through the optimized building information model analysis neural network, data analysis processing is performed on each of the multiple candidate building information models to obtain model identification data corresponding to each of the candidate building information models (that is, through network optimization processing, the optimized building information model analysis neural network can have the function of analyzing the model identification data corresponding to the candidate building information models).

[0083] For example, in some alternative implementations, the step of performing network optimization to form an optimized building information model analysis neural network may further include the following:

[0084] A first exemplary building information model data cluster is collected, and a network optimization process is performed on the first exemplary building information model data cluster to form a corresponding current building information model analysis neural network (exemplarily, the current building information model analysis neural network can be subjected to network optimization process based on an error between actual model identification data corresponding to the first exemplary building information model data included in the first exemplary building information model data cluster and predicted model identification data output by analysis, i.e., the included network parameters are optimized and adjusted, etc.), wherein the first exemplary building information model data in the first exemplary building information model data cluster is formed by performing actual model identification data configuration process on the second exemplary building information model data, and the second exemplary building information model data belongs to exemplary building information model data without actual model identification data screened out from the second exemplary building information model data cluster;

[0085] The first exemplary building information model data and other exemplary building information model data without actual model identification data are respectively loaded and processed to be loaded into the current building information model analysis neural network, and the building information model data description vectors corresponding to the first exemplary building information model data and the other exemplary building information model data without actual model identification data are analyzed and outputted, wherein the other exemplary building information model data without actual model identification data belongs to the exemplary building information model data without actual model identification data other than the second exemplary building information model data in the second exemplary building information model data cluster, and the building information model data description vectors are used to analyze and output corresponding prediction model identification data (exemplarily, the prediction unit included in the current building information model analysis neural network can be used to further process the building information model data description vectors to obtain the prediction model identification data, and the prediction unit can include a softmax function, etc.);

[0086] According to the building information model data description vectors respectively corresponding to the first exemplary building information model data and the other exemplary building information model data without actual model identification data, a target number of third exemplary building information model data are determined from the other exemplary building information model data without actual model identification data, and the third exemplary building information model data are marked as the supplementary second exemplary building information model data, and the data difference between the third exemplary building information model data and the current first exemplary building information model data cluster exceeds the data difference between the other exemplary building information model data without actual model identification data and the current first exemplary building information model data cluster;

[0087] According to the supplemented second exemplary building information model data, the current first exemplary building information model data cluster is supplemented, and, based on the current first exemplary building information model data cluster after the supplementation processing, the steps of collecting the current first exemplary building information model data cluster and performing network optimization processing according to the current first exemplary building information model data cluster to form a corresponding current building information model analysis neural network are performed, and, after the network optimization processing is completed, the current building information model analysis neural network is marked as an optimized building information model analysis neural network.

[0088] For example, in some alternative implementations, the building information model data description vector belongs to a description vector output by a key information mining operation in a data analysis process of the model identification data (exemplarily, the current building information model analysis neural network may further include a feature mining unit, and the feature mining unit may perform a key information mining operation on the exemplary building information model data to output a corresponding description vector. Specifically, feature mapping may be performed first to map to a corresponding feature space to form a corresponding initial description vector, and then a corresponding description vector may be determined based on the initial description vector. For example, the initial description vector may be directly determined as the corresponding description vector, or the initial description vector may be processed first to form a corresponding description vector, such as performing a significant feature analysis to obtain a corresponding description vector). Based on this, the step of determining a target number of third exemplary building information model data from the other exemplary building information model data without actual model identification data based on the building information model data description vectors corresponding to the first exemplary building information model data and the other exemplary building information model data without actual model identification data to mark as the supplementary second exemplary building information model data may further include the following contents:

[0089] A first building information model data description vector cluster and a second building information model data description vector cluster are extracted, wherein the first building information model data description vector in the first building information model data description vector cluster (which may refer to the first building information model data description vector cluster) belongs to the description vector output by the key information mining operation of the first exemplary building information model data, and the second building information model data description vector in the second building information model data description vector cluster (which may refer to the second building information model data description vector cluster) belongs to the description vector output by the key information mining operation of the other exemplary building information model data without actual model identification data;

[0090] Analyze and output target vector matching parameters (such as vector cosine similarity between description vectors) between each second building information model data description vector in the current second building information model data description vector cluster and the current first building information model data description vector cluster;

[0091] Filter out the second building information model data description vector corresponding to the target vector matching parameter with the maximum value, and mark other exemplary building information model data without actual model identification data corresponding to the second building information model data description vector as third exemplary building information model data;

[0092] According to the screened second building information model data description vector, the current first building information model data description vector cluster and the current second building information model data description vector cluster are adjusted to determine the third exemplary building information model data having a target number, the current first building information model data description vector cluster after the adjustment includes the screened second building information model data description vector, and the screened second building information model data description vector is marked as the first building information model data description vector, and the current second building information model data description vector cluster after the adjustment does not include the screened second building information model data description vector;

[0093] A marking process is performed on the target number of the third exemplary building information model data to mark them as supplementary second exemplary building information model data.

[0094] For example, in some alternative implementations, the step of analyzing and outputting a target vector matching parameter between each second building information model data description vector in the current second building information model data description vector cluster and the current first building information model data description vector cluster may further include the following:

[0095] For each second building information model data description vector in the current second building information model data description vector cluster, calculate a to-be-processed vector matching parameter between the second building information model data description vector and each first building information model data description vector in the current first building information model data description vector cluster (exemplarily, the to-be-processed vector matching parameter may refer to a vector cosine similarity between description vectors);

[0096] The to-be-processed vector matching parameter with the maximum value is marked to mark and form a target vector matching parameter between the second building information model data description vector and the first building information model data description vector cluster.

[0097] For example, in some alternative implementations, the step of acquiring the first exemplary building information model data cluster may further include the following:

[0098] A second exemplary building information model data cluster is collected, wherein the second exemplary building information model data cluster includes exemplary building information model data without actual model identification data corresponding to at least one configuration model identification data;

[0099] Respectively, from the exemplary building information model data without actual model identification data corresponding to each of the configuration model identification data, screen out the exemplary building information model data without actual model identification data for marking, so as to mark and form corresponding second exemplary building information model data, so as to form corresponding candidate second exemplary building information model data clusters, wherein the number of second exemplary building information model data included in the candidate second exemplary building information model data clusters is less than the number of exemplary building information model data without actual model identification data in the second exemplary building information model data clusters;

[0100] The second exemplary building information model data in the candidate second exemplary building information model data cluster is subjected to configuration processing of actual model identification data to form a corresponding current first exemplary building information model data cluster.

[0101] For example, in some alternative implementations, the step of acquiring the second exemplary building information model data cluster may further include the following:

[0102] An exemplary building information model data cluster is collected, wherein the exemplary building information model data cluster includes exemplary building information model data corresponding to at least one configuration model identification data;

[0103] Extracting multi-faceted model data content corresponding to each exemplary building information model data, wherein the multi-faceted model data content includes model component self data and model component relationship data of the exemplary building information model data (exemplarily, the model component self data can be used to reflect information of the corresponding model component itself, such as size, structure, etc., and the model component relationship data can be used to reflect the relationship between model components, such as connection relationship, position relationship, etc.);

[0104] The multi-level model data content corresponding to each exemplary building information model data is marked as exemplary building information model data without actual model identification data to form a corresponding second exemplary building information model data cluster.

[0105] For example, in some alternative implementations, the step of using the key information mining unit included in the target building quality analysis neural network to perform key information mining processing on the target building information model data to output a key information mining result representation vector corresponding to the target building information model data may further include the following contents:

[0106] Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing (i.e., embedding) to form a model data feature mapping vector corresponding to the target building information model data;

[0107] The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

[0108] For example, in some alternative implementations, the step of using the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector may further include the following contents:

[0109] Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector (as described later) to output the selected key information mining result representation vector corresponding to the first key information convolution subunit, the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit;

[0110] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit;

[0111] Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, at least one corresponding selected key information mining result representation vector is obtained.

[0112] For example, in some alternative implementations, for each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is convolved using the other key information convolution subunit to output the characterization vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and the characterization vector of the key information mining result to be processed and the output data of the previous key information convolution subunit are fused to output the step of the selected key information mining result characterization vector corresponding to the other key information convolution subunit, which may further include the following contents:

[0113] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit;

[0114] The representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit and the output data of the previous key information convolution subunit are concatenated to output the representation vector of the selected key information mining result corresponding to the other key information convolution subunit.

[0115] For example, in some alternative implementations, the step of performing a convolution operation on the output data of the previous key information convolution subunit using the other key information convolution subunit for each other key information convolution subunit other than the first key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit may further include the following contents:

[0116] For each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is filtered using the filter matrix included in the other key information convolution subunit (the matrix parameters of the filter matrix can be obtained by learning the corresponding exemplary data based on the configured initial parameters) to output the filtered key information mining result representation vector corresponding to the other key information convolution subunit;

[0117] Performing vector segmentation processing (which may be segmented into equal sizes) on the filtering key information mining result representation vector to form a local filtering key information mining result representation vector cluster corresponding to the filtering key information mining result representation vector, wherein the local filtering key information mining result representation vector cluster includes at least two local filtering key information mining result representation vectors;

[0118] For each of the local filtering key information mining result representation vectors, mark the local filtering key information mining result representation vector as a target local filtering key information mining result representation vector, and mark other local filtering key information mining result representation vectors other than the local filtering key information mining result representation vector as reference local filtering key information mining result representation vectors, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the target local filtering key information mining result representation vector, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector, and respectively perform association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector, and superimpose each of the association representation vectors to output the target association representation vector corresponding to the target local filtering key information mining result representation vector;

[0119] The target associated representation vector corresponding to each of the local filtering key information mining result representation vectors is merged (which is the inverse of the aforementioned segmentation process) with the key information mining result representation vector to be processed corresponding to the other key information convolution sub-units.

[0120] For example, in some alternative implementations, the step of performing association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and the value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector may further include the following contents:

[0121] The query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector are multiplied to output a corresponding association representation vector.

[0122] Combination Figure 3 The embodiment of the present invention further provides a monitoring system based on building informatization, which can be applied to the monitoring platform based on building informatization. The monitoring system based on building informatization can include the modules described below:

[0123] A data acquisition module is used to obtain the request information for obtaining the building information data to be processed, and parse the request information for obtaining the building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining the building information data to be processed, and, based on the building information model description data to be matched, determine a target building information model from multiple candidate building information models, and use the target building information model data corresponding to the target building information model as the data acquisition result corresponding to the request information for obtaining the building information data to be processed;

[0124] A key information mining module is used to perform key information mining processing on the target building information model data using the key information mining unit included in the target building quality analysis neural network to output a key information mining result representation vector corresponding to the target building information model data;

[0125] A key information extraction module, used to use the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector;

[0126] A building quality prediction module, for using the building quality prediction unit included in the target building quality analysis neural network to respectively perform building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors, so as to output at least two building quality prediction results corresponding to the target building information model data;

[0127] The prediction result fusion module is used to fuse the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model, and the target building quality prediction result is used to reflect the building quality status of the target building corresponding to the target building information model.

[0128] For example, in some alternative implementations, the key information mining module is specifically used to:

[0129] Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing to form a model data feature mapping vector corresponding to the target building information model data;

[0130] The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

[0131] For example, in some alternative implementations, the key information extraction module is specifically used to:

[0132] Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector to output a selected key information mining result representation vector corresponding to the first key information convolution subunit, wherein the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit;

[0133] For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit;

[0134] Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, at least one corresponding selected key information mining result representation vector is obtained.

[0135] In summary, the monitoring method and system based on building informationization provided by the present invention can utilize a key information mining unit to perform key information mining processing on target building information model data to output a key information mining result representation vector; utilize a key information extraction unit to perform at least one key information extraction processing on the key information mining result representation vector to output at least one selected key information mining result representation vector; utilize a building quality prediction unit to perform building quality prediction processing on the key information mining result representation vector and each selected key information mining result representation vector to output at least two building quality prediction results; perform fusion processing on at least two building quality prediction results to output a target building quality prediction result. Based on the aforementioned steps, since the key information mining result representation vector and each selected key information mining result representation vector are subjected to building quality prediction processing to output at least two building quality prediction results, and then fusion processing is performed to output the target building quality prediction result, the reliability of the target building quality prediction result is higher, thereby improving the reliability of building quality monitoring to a certain extent.

[0136] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A monitoring method based on building informatization, characterized in that: include: Acquire request information for obtaining building information data to be processed, and parse the request information for obtaining building information data to be processed to obtain building information model description data to be matched carried in the request information for obtaining building information data to be processed, and determine a target building information model from multiple candidate building information models based on the building information model description data to be matched, and use the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining building information data to be processed; Using the key information mining unit included in the target building quality analysis neural network, the target building information model data is subjected to key information mining processing to output a key information mining result representation vector corresponding to the target building information model data; Using the key information extraction unit included in the target building quality analysis neural network, perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector; Utilizing the building quality prediction unit included in the target building quality analysis neural network, respectively performing building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors, so as to output at least two building quality prediction results corresponding to the target building information model data; Performing fusion processing on the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model, wherein the target building quality prediction result is used to reflect the building quality of the target building corresponding to the target building information model; The step of using the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector includes: Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector to output a selected key information mining result representation vector corresponding to the first key information convolution subunit, wherein the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit; For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit; Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, obtaining at least one corresponding selected key information mining result representation vector; The step of performing a convolution operation on the output data of the previous key information convolution subunit using the other key information convolution subunit for each other key information convolution subunit other than the first key information convolution subunit to output a characterization vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fusing the characterization vector of the key information mining result to be processed with the output data of the previous key information convolution subunit to output a characterization vector of the selected key information mining result corresponding to the other key information convolution subunit includes: For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit; The representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit is concatenated with the output data of the previous key information convolution subunit to output the representation vector of the selected key information mining result corresponding to the other key information convolution subunit; The step of performing a convolution operation on the output data of the previous key information convolution subunit using each other key information convolution subunit other than the first key information convolution subunit to output a characterization vector of the key information mining result to be processed corresponding to the other key information convolution subunit includes: For each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is filtered using the filter matrix included in the other key information convolution subunit to output a filtered key information mining result representation vector corresponding to the other key information convolution subunit; Performing vector segmentation processing on the filtering key information mining result representation vector to form a local filtering key information mining result representation vector cluster corresponding to the filtering key information mining result representation vector, wherein the local filtering key information mining result representation vector cluster includes at least two local filtering key information mining result representation vectors; For each of the local filtering key information mining result representation vectors, mark the local filtering key information mining result representation vector as a target local filtering key information mining result representation vector, and mark other local filtering key information mining result representation vectors other than the local filtering key information mining result representation vector as reference local filtering key information mining result representation vectors, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the target local filtering key information mining result representation vector, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector, and respectively perform association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector, and superimpose each of the association representation vectors to output the target association representation vector corresponding to the target local filtering key information mining result representation vector; The target association representation vector corresponding to each of the local filtering key information mining result representation vectors is merged to output the key information mining result representation vectors to be processed corresponding to the other key information convolution sub-units.

2. The monitoring method based on building informatization according to claim 1, characterized in that: The step of using the key information mining unit included in the target building quality analysis neural network to perform key information mining processing on the target building information model data to output a key information mining result representation vector corresponding to the target building information model data includes: Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing to form a model data feature mapping vector corresponding to the target building information model data; The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

3. The monitoring method based on building informatization according to claim 1, characterized in that: The step of respectively performing association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and the value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector includes: The query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector are multiplied to output the corresponding association representation vector.

4. The monitoring method based on building informatization according to any one of claims 1 to 3, characterized in that: The step of obtaining the request information for obtaining the building information data to be processed, parsing the request information for obtaining the building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining the building information data to be processed, and determining a target building information model from a plurality of candidate building information models based on the building information model description data to be matched, and using the target building information model data corresponding to the target building information model as the data collection result corresponding to the request information for obtaining the building information data to be processed, comprises: Acquire the request information for obtaining the building information data to be processed, and parse the request information for obtaining the building information data to be processed to obtain the building information model description data to be matched carried in the request information for obtaining the building information data to be processed; Acquire model identification data corresponding to each candidate building information model among a plurality of candidate building information models, wherein the model identification data is formed in advance by performing data analysis processing on the corresponding candidate building information model to identify the corresponding candidate building information model; Performing data matching calculations on the to-be-matched building information model description data and the model identification data corresponding to each of the candidate building information models, so as to output data matching parameters between the to-be-matched building information model description data and each of the candidate building information models; According to the data matching parameters between the building information model description data to be matched and each of the candidate building information models, a target building information model is determined from the multiple candidate building information models, and the target building information model data corresponding to the target building information model is used as the data collection result corresponding to the request information for obtaining the building information data to be processed.

5. A monitoring system based on building informatization, characterized in that: include: A data acquisition module is used to obtain a request for obtaining building information data to be processed, and parse the request for obtaining building information data to be processed to obtain the building information model description data to be matched carried in the request for obtaining building information data to be processed, and, based on the building information model description data to be matched, determine a target building information model from multiple candidate building information models, and use the target building information model data corresponding to the target building information model as the data acquisition result corresponding to the request for obtaining building information data to be processed; A key information mining module is used to perform key information mining processing on the target building information model data using the key information mining unit included in the target building quality analysis neural network to output a key information mining result representation vector corresponding to the target building information model data; A key information extraction module, used to use the key information extraction unit included in the target building quality analysis neural network to perform at least one key information extraction process on the key information mining result representation vector to output at least one corresponding selected key information mining result representation vector; A building quality prediction module, for using the building quality prediction unit included in the target building quality analysis neural network to respectively perform building quality prediction processing on the key information mining result representation vector and each of the selected key information mining result representation vectors, so as to output at least two building quality prediction results corresponding to the target building information model data; A prediction result fusion module, used for fusing the at least two building quality prediction results to output a target building quality prediction result corresponding to the target building information model, wherein the target building quality prediction result is used to reflect the building quality of the target building corresponding to the target building information model; The key information extraction module is specifically used for: Using the first key information convolution subunit in the key information extraction unit included in the target building quality analysis neural network, a convolution operation is performed on the key information mining result representation vector to output a selected key information mining result representation vector corresponding to the first key information convolution subunit, wherein the key information extraction unit includes a plurality of key information convolution subunits, and the plurality of key information convolution subunits are sequentially connected so that the output data of the previous key information convolution subunit is used as the input data of the next key information convolution subunit; For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and fuse the representation vector of the key information mining result to be processed and the output data of the previous key information convolution subunit to output a representation vector of the selected key information mining result corresponding to the other key information convolution subunit; Based on the selected key information mining result representation vector corresponding to each of the key information convolution subunits, obtaining at least one corresponding selected key information mining result representation vector; For each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is convolved by the other key information convolution subunit to output a characterization vector of the key information mining result to be processed corresponding to the other key information convolution subunit, and the characterization vector of the key information mining result to be processed and the output data of the previous key information convolution subunit are fused to output a characterization vector of the selected key information mining result corresponding to the other key information convolution subunit, including: For each other key information convolution subunit other than the first key information convolution subunit, use the other key information convolution subunit to perform a convolution operation on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit; The representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit is concatenated with the output data of the previous key information convolution subunit to output the representation vector of the selected key information mining result corresponding to the other key information convolution subunit; For each other key information convolution subunit other than the first key information convolution subunit, using the other key information convolution subunit, a convolution operation is performed on the output data of the previous key information convolution subunit to output a representation vector of the key information mining result to be processed corresponding to the other key information convolution subunit, including: For each other key information convolution subunit other than the first key information convolution subunit, the output data of the previous key information convolution subunit is filtered using the filter matrix included in the other key information convolution subunit to output a filtered key information mining result representation vector corresponding to the other key information convolution subunit; Performing vector segmentation processing on the filtering key information mining result representation vector to form a local filtering key information mining result representation vector cluster corresponding to the filtering key information mining result representation vector, wherein the local filtering key information mining result representation vector cluster includes at least two local filtering key information mining result representation vectors; For each of the local filtering key information mining result representation vectors, mark the local filtering key information mining result representation vector as a target local filtering key information mining result representation vector, and mark other local filtering key information mining result representation vectors other than the local filtering key information mining result representation vector as reference local filtering key information mining result representation vectors, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the target local filtering key information mining result representation vector, and analyze the query distribution matrix, key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector, and respectively perform association analysis on the query distribution matrix corresponding to the target local filtering key information mining result representation vector and the key distribution matrix and value distribution matrix corresponding to the reference local filtering key information mining result representation vector to output each corresponding association representation vector, and superimpose each of the association representation vectors to output the target association representation vector corresponding to the target local filtering key information mining result representation vector; The target association representation vector corresponding to each of the local filtering key information mining result representation vectors is merged to output the key information mining result representation vectors to be processed corresponding to the other key information convolution sub-units.

6. The monitoring system based on building informationization as claimed in claim 5, characterized in that: The key information mining module is specifically used for: Using a feature mapping subunit in a key information mining unit included in a target building quality analysis neural network, the target building information model data is subjected to feature space mapping processing to form a model data feature mapping vector corresponding to the target building information model data; The vector encoding subunit in the key information mining unit included in the target building quality analysis neural network is used to perform vector encoding processing on the model data feature mapping vector to form a key information mining result representation vector corresponding to the target building information model data.

Images