Data processing method and system based on smart building

Abstract

The data processing method and system based on smart buildings provided by the present invention relate to the field of artificial intelligence technology. In the present invention, the user behavior analysis operation is performed on the sequence of user behavior monitoring images to be processed by using an optimized user behavior analysis neural network, and the optimized user behavior analysis neural network performs network optimization processing on the second user behavior analysis neural network based on multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis basis includes the output data of the second user behavior analysis neural network and the first user behavior analysis neural network, and the first user behavior analysis neural network belongs to a neural network that has been subjected to network optimization processing; the user behavior demand characterization data corresponding to the sequence of user behavior monitoring images to be processed analyzed by the optimized user behavior analysis neural network is output. Based on the above content, the problem of poor reliability of user behavior data analysis in the prior art can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a data processing method and system based on smart buildings. Background Art

[0002] Smart buildings refer to the optimization of the building structure, system, service and management according to the needs of users, so as to provide users with an efficient, comfortable and convenient humanized building environment. In the process of realizing smart buildings, data collection is generally performed on building users, such as user behavior collection. Specifically, user behavior data can be collected through video monitoring and other operations. Then, the user's behavior data can be analyzed to determine the user's needs. In this way, corresponding building smart management and control can be carried out based on the needs. However, in the existing technology, there is a problem of poor reliability in analyzing user behavior data. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a data processing method and system based on smart buildings to improve the problem of poor reliability of user behavior data analysis in the prior art.

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

[0005] A data processing method based on smart buildings, comprising:

[0006] A sequence of user behavior monitoring images to be processed corresponding to a target smart building is collected, wherein the sequence of user behavior monitoring images to be processed includes multiple frames of user behavior monitoring images to be processed, and the multiple frames of user behavior monitoring images to be processed are formed based on image collection of at least one monitored building area of ​​the target smart building;

[0007] The sequence of user behavior monitoring images to be processed is loaded to be loaded into the optimized user behavior analysis neural network, and the optimized user behavior analysis neural network is used to perform a user behavior analysis operation on the sequence of user behavior monitoring images to be processed, wherein the optimized user behavior analysis neural network performs network optimization processing on a second user behavior analysis neural network according to a multi-dimensional error index, and in the process of analyzing the error index, the analysis basis includes output data of the second user behavior analysis neural network and output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to a user behavior analysis neural network that has been subjected to network optimization processing, and the second user behavior analysis neural network belongs to a user behavior analysis neural network that has not been subjected to network optimization processing;

[0008] Output the user behavior demand representation data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network, wherein the user behavior demand representation data is used to reflect the demand information of the building users in the target smart building.

[0009] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the network optimization processing of optimizing the user behavior analysis neural network includes:

[0010] An exemplary image sequence cluster is extracted, the exemplary image sequence cluster includes an exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data corresponding to the first user behavior analysis neural network, the exemplary image sequence cluster also includes an exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data, the exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data is used for network optimization processing to form the first user behavior analysis neural network, and the exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data has not been used for network optimization processing to form the first user behavior analysis neural network;

[0011] Using the first user behavior analysis neural network, mining a high-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster;

[0012] Using a second user behavior analysis neural network, mining a low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster, and performing a user behavior analysis operation based on each of the low-precision behavior data description vectors, and then calculating and outputting a corresponding behavior demand analysis error index based on each of the output low-precision user behavior demand representation data;

[0013] Performing error index analysis based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output a corresponding data internal mining error index;

[0014] Perform error index analysis based on low-precision behavior data description vectors corresponding to each two of the exemplary user behavior monitoring image sequences having different behavior demand representation data, so as to output corresponding data-to-data mining error index;

[0015] Based on the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index, the second user behavior analysis neural network is subjected to network optimization processing to form a corresponding optimized user behavior analysis neural network.

[0016] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the step of performing error index analysis based on the low-precision behavior data description vectors corresponding to each two of the exemplary user behavior monitoring image sequences having different behavior demand representation data to output the corresponding data mining error index includes:

[0017] Combining each of the exemplary user behavior monitoring image sequences and each of the exemplary user behavior monitoring image sequences in the exemplary image sequence cluster to form a corresponding exemplary user behavior monitoring image sequence combination, so as to form a corresponding plurality of exemplary user behavior monitoring image sequence combinations;

[0018] Determine the vector correlation coefficient according to the low-precision behavior data description vector corresponding to each of the exemplary user behavior monitoring image sequence combinations;

[0019] Analyzing the local data mining error index corresponding to each of the exemplary user behavior monitoring image sequence combinations according to the determined vector correlation coefficient;

[0020] The local data-to-data mining error indicators corresponding to each first exemplary user behavior monitoring image sequence combination are fused to output the corresponding data-to-data mining error indicators, wherein the two exemplary user behavior monitoring image sequences included in the first exemplary user behavior monitoring image sequence combination have different behavior demand characterization data.

[0021] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, before the step of performing network optimization processing on the second user behavior analysis neural network based on the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form a corresponding optimized user behavior analysis neural network, the network optimization processing of the optimized user behavior analysis neural network further includes:

[0022] Perform error index analysis based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to each two exemplary user behavior monitoring image sequences with different behavior demand representation data, so as to output the corresponding inter-data hybrid mining error index;

[0023] The step of performing network optimization processing on the second user behavior analysis neural network according to the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index to form a corresponding step of optimizing the user behavior analysis neural network includes:

[0024] The internal data mining error index, the inter-data mining error index, the inter-data mixed mining error index and the behavior demand analysis error index are fused to output a corresponding fusion error index; and, based on the fusion error index, the second user behavior analysis neural network is optimized to form a corresponding optimized user behavior analysis neural network.

[0025] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the step of performing error index analysis on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output the corresponding data internal mining error index includes:

[0026] Determine a vector correlation coefficient according to the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence, so as to output the corresponding vector correlation coefficient;

[0027] Analyzing the local data mining error index corresponding to each exemplary user behavior monitoring image sequence according to the determined vector correlation coefficient, wherein there is a non-positive object relationship between the local data mining error index corresponding to the exemplary user behavior monitoring image sequence and the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence;

[0028] The local data inter-mining error indicators corresponding to each exemplary user behavior monitoring image sequence are fused to output the corresponding data internal mining error indicators.

[0029] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the deterministic user behavior demand representation data includes a variety of different user behavior demand representation data, and the step of extracting an exemplary image sequence cluster includes:

[0030] Extracting an original exemplary image sequence cluster, the original exemplary image sequence cluster includes a central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data, and the projection coordinates of the central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data in the description vector distribution domain corresponding to the deterministic user behavior demand characterization data are associated with the projection coordinates of all exemplary image sequences corresponding to the deterministic user behavior demand characterization data in the description vector distribution domain;

[0031] An extended exemplary user behavior monitoring image sequence cluster corresponding to the extended user behavior demand characterization data is extracted, and an exemplary image sequence cluster is formed according to the extended exemplary user behavior monitoring image sequence cluster and the original exemplary image sequence cluster.

[0032] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, before the step of extracting the original exemplary image sequence cluster, the network optimization process of optimizing the user behavior analysis neural network further includes:

[0033] Extracting all exemplary image sequences corresponding to each deterministic user behavior demand representation data, wherein all exemplary image sequences include an optimized exemplary user behavior monitoring image sequence corresponding to the first user behavior analysis neural network;

[0034] For all exemplary image sequences corresponding to each of the deterministic user behavior demand characterization data, classify the behavior data description vectors of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand characterization data;

[0035] In each of the data classification sets, the behavior data description vectors that match the corresponding classification center behavior data description vectors are determined respectively, and the optimized exemplary user behavior monitoring image sequence corresponding to the determined behavior data description vectors is marked as the center exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data.

[0036] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the step of classifying the behavior data description vector of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand representation data includes:

[0037] Extracting the number of classification indicators corresponding to the deterministic user behavior demand representation data;

[0038] The behavior data description vector corresponding to each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences is classified according to the number of classification indicators to form a plurality of data classification sets whose number corresponding to the deterministic user behavior demand characterization data is consistent with the number of classification indicators.

[0039] In some preferred embodiments, in the above-mentioned data processing method based on smart buildings, the number of the extracted extensible user behavior demand representation data is greater than 1, and after forming the optimized user behavior analysis neural network, it also includes:

[0040] Counting the number of demand types corresponding to the extended user behavior demand representation data, and discarding a plurality of central exemplary user behavior monitoring image sequences in the original exemplary image sequence cluster according to the number of demand types;

[0041] Analyzing a central exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data from the extended exemplary user behavior monitoring image sequence cluster corresponding to the extended user behavior demand characterization data;

[0042] The central exemplary user behavior monitoring image sequence corresponding to the extensible user behavior demand characterization data is saved.

[0043] An embodiment of the present invention also provides a data processing system based on a smart building, including a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the above-mentioned data processing method based on the smart building.

[0044] The data processing method and system based on smart buildings provided by the embodiment of the present invention can collect the to-be-processed user behavior monitoring image sequence corresponding to the target smart building; use the optimized user behavior analysis neural network to perform user behavior analysis operations on the to-be-processed user behavior monitoring image sequence; and output the user behavior demand characterization data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network. Based on this, since the optimized user behavior analysis neural network performs network optimization processing on the second user behavior analysis neural network based on multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis basis includes the output data of the second user behavior analysis neural network and the output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to the user behavior analysis neural network that has been processed by network optimization. In this way, the optimized user behavior analysis neural network can not only learn based on the second user behavior analysis neural network, but also learn the information learned by the first user behavior analysis neural network in the process of network optimization processing, so that the analysis reliability of the optimized user behavior analysis neural network is higher, thereby improving the problem of poor reliability of user behavior data analysis in the prior art.

[0045] 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

[0046] Figure 1 A structural block diagram of a data processing system based on a smart building provided in an embodiment of the present invention.

[0047] Figure 2 A flowchart of the steps of the data processing method based on smart buildings provided in an embodiment of the present invention.

[0048] Figure 3 A schematic diagram of various modules included in the data processing device based on smart buildings provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] 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.

[0050] 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.

[0051] like Figure 1 As shown, an embodiment of the present invention provides a data processing system based on a smart building. The data processing system based on a smart building may include a memory and a processor.

[0052] 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 data processing method based on smart buildings provided by the embodiment of the present invention.

[0053] It can be selected that in some feasible implementations, the memory can be, but not limited to, random access memory (RAM), read only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable read-only memory (EEPROM), etc. The processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), a system on chip (SoC), etc.; it can 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.

[0054] Optionally, in some feasible implementations, the smart building-based data processing system may be a server with data processing capabilities.

[0055] Combination Figure 2 The embodiment of the present invention further provides a data processing method based on smart buildings, which can be applied to the data processing system based on smart buildings. The method steps defined in the process related to the data processing method based on smart buildings can be implemented by the data processing system based on smart buildings. Figure 2 The specific process shown is explained in detail.

[0056] Step S110, collecting a sequence of user behavior monitoring images to be processed corresponding to the target smart building.

[0057] In an embodiment of the present invention, the data processing system based on smart buildings can collect a sequence of user behavior monitoring images to be processed corresponding to the target smart building (such as obtained based on the transmission of the monitoring terminal device). The sequence of user behavior monitoring images to be processed includes multiple frames of user behavior monitoring images to be processed, and the multiple frames of user behavior monitoring images to be processed are formed based on image collection of at least one monitored building area of ​​the target smart building.

[0058] Step S120, loading the to-be-processed user behavior monitoring image sequence to the optimized user behavior analysis neural network, and performing a user behavior analysis operation on the to-be-processed user behavior monitoring image sequence using the optimized user behavior analysis neural network.

[0059] In an embodiment of the present invention, the data processing system based on smart buildings can load the user behavior monitoring image sequence to be processed to load it into the optimized user behavior analysis neural network, and use the optimized user behavior analysis neural network to perform user behavior analysis operations on the user behavior monitoring image sequence to be processed, that is, analyze the corresponding behavior needs. The optimized user behavior analysis neural network is formed by performing network optimization processing on the second user behavior analysis neural network based on multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis basis includes the output data of the second user behavior analysis neural network and the output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to the user behavior analysis neural network that has been processed by network optimization, and the second user behavior analysis neural network belongs to the user behavior analysis neural network that has not been processed by network optimization (exemplarily, performing user behavior analysis operations on the user behavior monitoring image sequence to be processed can be either directly analyzing the user behavior monitoring image sequence to be processed, or first performing behavior action recognition processing on the user behavior monitoring image sequence to be processed, and then analyzing based on the obtained behavior action sequence, etc.).

[0060] Step S130, outputting the user behavior demand representation data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network.

[0061] In an embodiment of the present invention, the data processing system based on smart buildings can output user behavior demand representation data corresponding to the sequence of user behavior monitoring images to be processed analyzed by the optimized user behavior analysis neural network. The user behavior demand representation data is used to reflect the demand information of building users in the target smart building.

[0062] Based on the above method, namely including steps S110-S130, since the optimized user behavior analysis neural network is formed by performing network optimization processing on the second user behavior analysis neural network based on multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis is based on the output data of the second user behavior analysis neural network and the output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to the user behavior analysis neural network that has been processed by network optimization, so the optimized user behavior analysis neural network can not only learn based on the second user behavior analysis neural network, but also learn the information learned by the first user behavior analysis neural network in the process of network optimization processing, so that the analysis reliability of the optimized user behavior analysis neural network is higher, thereby improving the problem of poor reliability of user behavior data analysis in the prior art.

[0063] Optionally, in some feasible implementations, the network optimization process of optimizing the user behavior analysis neural network may further include the following details:

[0064] An exemplary image sequence cluster is extracted, the exemplary image sequence cluster includes an exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand representation data corresponding to the first user behavior analysis neural network, the exemplary image sequence cluster also includes an exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand representation data, the exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand representation data is used for network optimization processing to form the first user behavior analysis neural network, and the exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand representation data has not been used for network optimization processing to form the first user behavior analysis neural network (in this way, new and old integrated learning of exemplary data can be achieved);

[0065] Using the first user behavior analysis neural network, a high-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster is mined (since the first user behavior analysis neural network has been processed by network optimization, the accuracy of the mined behavior data description vector is relatively high, wherein the mining of the behavior data description vector may include projection of a description vector distribution domain (such as feature space), that is, projecting the exemplary user behavior monitoring image into the description vector distribution domain to form a corresponding behavior data description vector, or, after the projection, further screening or filtering may be performed);

[0066] Using the second user behavior analysis neural network, mining the low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster (since the second user behavior analysis neural network has not been subjected to network optimization processing, the precision of the mined behavior data description vector is relatively low), and performing a user behavior analysis operation according to each of the low-precision behavior data description vectors, and then calculating and outputting a corresponding behavior demand analysis error index according to each of the output low-precision user behavior demand representation data (exemplarily, the corresponding behavior demand analysis error index can be determined according to the difference between each of the output low-precision user behavior demand representation data and the actual user behavior demand representation data);

[0067] Performing error index analysis based on the high-precision behavior data description vector and the low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output a corresponding data internal mining error index (exemplarily, for exemplary user behavior monitoring image sequence A, the difference between the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence A and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence A can be calculated to determine the corresponding data internal mining error index);

[0068] Performing error index analysis based on the low-precision behavior data description vectors corresponding to each two of the exemplary user behavior monitoring image sequences having different behavior demand representation data, so as to output a corresponding data mining error index (exemplarily, the exemplary user behavior monitoring image sequence A and the exemplary user behavior monitoring image sequence B have different behavior demand representation data, and therefore, the difference between the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence A and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence B can be calculated to obtain the corresponding data mining error index);

[0069] Based on the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index, the second user behavior analysis neural network is subjected to network optimization processing to form a corresponding optimized user behavior analysis neural network (exemplarily, the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index can be fused by weighted summation, and then network optimization processing is performed based on the obtained fused error index to form a corresponding optimized user behavior analysis neural network).

[0070] Alternatively, in some feasible implementations, the step of performing error index analysis based on the low-precision behavior data description vectors corresponding to each two exemplary user behavior monitoring image sequences having different behavior demand representation data to output the corresponding data mining error index may further include the following details:

[0071] Combining each of the exemplary user behavior monitoring image sequences and each of the exemplary user behavior monitoring image sequences in the exemplary image sequence cluster to form a corresponding exemplary user behavior monitoring image sequence combination, so as to form a corresponding plurality of exemplary user behavior monitoring image sequence combinations (i.e., combining the exemplary user behavior monitoring image sequences in the exemplary image sequence cluster in pairs to form a plurality of exemplary user behavior monitoring image sequence combinations);

[0072] Determine the vector correlation coefficients according to the low-precision behavior data description vectors corresponding to each of the exemplary user behavior monitoring image sequence combinations (exemplarily, for each of the exemplary user behavior monitoring image sequence combinations, the cosine similarity calculation can be performed on the low-precision behavior data description vectors corresponding to the two exemplary user behavior monitoring image sequences included in the exemplary user behavior monitoring image sequence combination to obtain the corresponding vector correlation coefficients);

[0073] Analyzing the local data mining error index corresponding to each of the exemplary user behavior monitoring image sequence combinations according to the determined vector correlation coefficient;

[0074] The local data-to-data mining error indicators corresponding to each first exemplary user behavior monitoring image sequence combination are fused to output the corresponding data-to-data mining error indicators, wherein the two exemplary user behavior monitoring image sequences included in the first exemplary user behavior monitoring image sequence combination have different behavior demand characterization data.

[0075] Wherein, it can be selected that, in some feasible implementation manners, the step of analyzing the local data mining error index corresponding to each of the exemplary user behavior monitoring image sequence combinations according to the determined vector correlation coefficient can further include the following details:

[0076] For each of the exemplary user behavior monitoring image sequence combinations, performing a difference operation on the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence combination and a pre-configured reference vector correlation coefficient (e.g., subtracting the former from the latter, and the specific value of the reference vector correlation coefficient is not limited) to output a corresponding difference vector correlation coefficient;

[0077] If the vector correlation coefficient is smaller than the reference vector correlation coefficient, the pre-configured reference mining error indicator is marked as the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence, and the reference mining error indicator is smaller than the difference vector correlation coefficient (exemplarily, the reference mining error indicator being smaller than the difference vector correlation coefficient may mean that the reference mining error indicator is smaller than each of the difference vector correlation coefficients that may be calculated; in addition, marking the reference mining error indicator may mean configuring the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence as the reference mining error indicator, so that the second user behavior analysis neural network may perform network optimization along the direction of marking the reference mining error indicator as the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence, so that the vector correlation coefficient between the exemplary user behavior monitoring image sequences of different behavior demand representation data may be smaller than the reference vector correlation coefficient, so that each mined description vector and the description vector corresponding to other behavior demand representation data can be easily distinguished and identified);

[0078] If the vector correlation coefficient is not less than the reference vector correlation coefficient, the difference vector correlation coefficient is marked as the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence (that is, the difference vector correlation coefficient can be used as the corresponding local data mining error indicator).

[0079] Wherein, it can be selected that, in some feasible implementation manners, the step of fusing the local data mining error indicators corresponding to each first exemplary user behavior monitoring image sequence combination to output the corresponding data mining error indicators can further include the following details:

[0080] For each of the exemplary user behavior monitoring image sequence combinations, actual user behavior demand characterization data corresponding to the exemplary user behavior monitoring image sequence included in the exemplary user behavior monitoring image sequence combination is extracted, and based on the actual user behavior demand characterization data, demand characterization parameters corresponding to the exemplary user behavior monitoring image sequence combination are analyzed, the demand characterization parameters being determined based on target analysis rules, if the actual user behavior demand characterization data corresponding to the exemplary user behavior monitoring image sequences included in the exemplary user behavior monitoring image sequence combination are different, then the demand characterization parameters analyzed based on the target analysis rules are equal to the first configuration parameters, if the actual user behavior demand characterization data corresponding to the exemplary user behavior monitoring image sequences included in the exemplary user behavior monitoring image sequence combination are the same, then the demand characterization parameters analyzed based on the target analysis rules are equal to the second configuration parameters (the specific values ​​of the first configuration parameter and the second configuration parameter are not limited, as long as they are different);

[0081] If the demand characterization parameter is equal to the first configuration parameter, the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence combination is marked to form the first local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence combination; if the demand characterization parameter is equal to the second configuration parameter, the local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence combination is replaced with a zero value to form the first local data mining error indicator corresponding to the exemplary user behavior monitoring image sequence combination;

[0082] The first local data mining error indicators corresponding to each of the exemplary user behavior monitoring image sequence combinations are accumulated to form a corresponding data mining error indicator (exemplarily, the first local data mining error indicators corresponding to each of the exemplary user behavior monitoring image sequence combinations can be superimposed to obtain a corresponding data mining error indicator).

[0083] Optionally, in some feasible implementations, before the step of performing network optimization processing on the second user behavior analysis neural network based on the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form a corresponding optimized user behavior analysis neural network, the network optimization processing of the optimized user behavior analysis neural network may further include the following detailed contents:

[0084] An error index analysis is performed based on the high-precision behavior data description vectors and the corresponding low-precision behavior data description vectors corresponding to each two exemplary user behavior monitoring image sequences having different behavior demand representation data, so as to output the corresponding data mixed mining error index (for example, an exemplary user behavior monitoring image sequence combination includes an exemplary user behavior monitoring image sequence m and an exemplary user behavior monitoring image sequence n, so that the high-precision behavior data description vector and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence combination can be the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, which can be based on the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m The error index analysis is performed on the low-precision behavior data description vector and the exemplary user behavior monitoring image sequence n corresponding to the exemplary user behavior monitoring image sequence to obtain the local data mining error index of the exemplary user behavior monitoring image sequence combination; or the high-precision behavior data description vector and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence combination may be the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, and the error index analysis may be performed based on the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n to obtain the local data mining error index of the exemplary user behavior monitoring image sequence combination;Alternatively, the high-precision behavior data description vector and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence combination may include the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, as well as the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, and an error index analysis may be performed based on the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, respectively, to obtain a local data interval. Mining error indicators, and, according to the low-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence m and the high-precision behavior data description vector corresponding to the exemplary user behavior monitoring image sequence n, performing error indicator analysis to obtain a local data mining error indicator, then calculating the mean of the two local data mining error indicators to obtain the local data mining error indicator of the exemplary user behavior monitoring image sequence combination, and finally, superimposing the local data mining error indicators of all the exemplary user behavior monitoring image sequence combinations to obtain the corresponding data mixed mining error indicator; in addition, the corresponding local data mining error indicator can be analyzed based on the corresponding vector correlation coefficient, as described above);

[0085] The step of performing network optimization processing on the second user behavior analysis neural network according to the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index to form a corresponding step of optimizing the user behavior analysis neural network includes:

[0086] The internal data mining error index, the inter-data mining error index, the inter-data mixed mining error index and the behavior demand analysis error index are fused (such as weighted summation processing to achieve fusion processing) to output a corresponding fusion error index; and, based on the fusion error index, the second user behavior analysis neural network is optimized to form a corresponding optimized user behavior analysis neural network.

[0087] Alternatively, in some feasible implementations, the step of performing error index analysis based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output the corresponding data internal mining error index may further include the following details:

[0088] Determine a vector correlation coefficient based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence to output the corresponding vector correlation coefficient; analyze the local data mining error index corresponding to each exemplary user behavior monitoring image sequence based on the determined vector correlation coefficient, and there is a non-positive object relationship between the local data mining error index corresponding to the exemplary user behavior monitoring image sequence and the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence (that is, the larger the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence, the smaller the local data mining error index corresponding to the exemplary user behavior monitoring image sequence can be; the smaller the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence, the larger the local data mining error index corresponding to the exemplary user behavior monitoring image sequence can be);

[0089] The local data-inter-mining error indicators corresponding to each exemplary user behavior monitoring image sequence are fused (such as superposition operations) to output the corresponding data-internal mining error indicators.

[0090] Optionally, in some feasible implementations, the deterministic user behavior demand characterization data includes a variety of different user behavior demand characterization data. Based on this, the step of extracting an exemplary image sequence cluster may further include the following details:

[0091] Extracting an original exemplary image sequence cluster, the original exemplary image sequence cluster includes a central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data, and the projection coordinates of the central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data in a description vector distribution domain corresponding to the deterministic user behavior demand characterization data are associated with the projection coordinates of all exemplary image sequences corresponding to the deterministic user behavior demand characterization data in the description vector distribution domain (e.g., similar, exemplarily, the description vector distribution domain may refer to a space formed by or in which the corresponding description vectors are located);

[0092] An extended exemplary user behavior monitoring image sequence cluster corresponding to the extended user behavior demand characterization data is extracted, and an exemplary image sequence cluster is formed based on the extended exemplary user behavior monitoring image sequence cluster and the original exemplary image sequence cluster (exemplarily, the extended exemplary user behavior monitoring image sequence cluster and the original exemplary image sequence cluster can be merged).

[0093] Optionally, in some feasible implementations, before the step of extracting the original exemplary image sequence cluster, the network optimization process of optimizing the user behavior analysis neural network may further include the following details:

[0094] Extracting all exemplary image sequences corresponding to each deterministic user behavior demand representation data, wherein all exemplary image sequences include an optimized exemplary user behavior monitoring image sequence corresponding to the first user behavior analysis neural network (i.e., an exemplary user behavior monitoring image sequence used in the network optimization processing of the first user behavior analysis neural network);

[0095] For all exemplary image sequences corresponding to each of the deterministic user behavior demand characterization data, classify the behavior data description vectors of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand characterization data;

[0096] In each of the data classification sets, a behavior data description vector that matches the corresponding classification center behavior data description vector is determined respectively (exemplarily, the classification center behavior data description vector corresponding to the data classification set may refer to the average value of the behavior data description vectors of each optimized used exemplary user behavior monitoring image sequence included in the data classification set, and the behavior data description vector that matches the corresponding classification center behavior data description vector may refer to the behavior data description vector with the smallest difference with the classification center behavior data description vector), and the optimized used exemplary user behavior monitoring image sequence corresponding to the determined behavior data description vector is marked as the center exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data (exemplarily, for one of the deterministic user behavior demand characterization data, it may have multiple data classification sets, and therefore, multiple center exemplary user behavior monitoring image sequences may be determined accordingly).

[0097] Alternatively, in some feasible implementations, the step of classifying the behavior data description vector of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand representation data may further include the following details:

[0098] Extracting the number of classification indicators corresponding to the deterministic user behavior demand representation data (the number of classification indicators is used to reflect the number of demand types of the deterministic user behavior demand representation data);

[0099] Each of the optimized behavior data description vectors corresponding to the exemplary user behavior monitoring image sequences in all the exemplary image sequences is classified according to the number of classification indicators to form multiple data classification sets whose number corresponding to the deterministic user behavior demand characterization data is consistent with the number of classification indicators (the specific classification algorithm is not limited).

[0100] It can be selected that, in some feasible implementations, the number of the extracted extensible user behavior demand representation data is greater than 1, that is, a plurality of the extensible user behavior demand representation data are extracted. Based on this, after forming the optimized user behavior analysis neural network, the data processing method based on smart buildings can further include the following details:

[0101] Count the number of demand types corresponding to the extended user behavior demand characterization data, and, based on the number of demand types, discard multiple central exemplary user behavior monitoring image sequences in the original exemplary image sequence cluster; analyze the central exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data in the extended exemplary user behavior demand characterization data cluster; save the central exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data (in this way, while keeping the overall data volume of the stored exemplary data stable, the exemplary data can be effectively updated, so that in subsequent learning optimization, it can better match the current actual situation, that is, the new optimized user behavior analysis neural network formed based on the subsequent learning optimization has higher reliability for the current behavior demand analysis and is more in line with the actual situation).

[0102] Wherein, it can be selected that, in some feasible implementation manners, the step of discarding a plurality of central exemplary user behavior monitoring image sequences in the original exemplary image sequence cluster according to the number of demand types may further include the following details:

[0103] According to the number of demand types, the number of discarding indicators of the exemplary user behavior monitoring image sequence to be discarded corresponding to each deterministic user behavior demand representation data is analyzed (exemplarily, the number of demand types can be used as the corresponding number of discarding indicators, or a positive correlation value of the number of demand types can be used as the corresponding number of discarding indicators);

[0104] For each of the deterministic user behavior demand characterization data, according to the optimized and used exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data, analyze the distribution domain center data description vector of the description vector distribution domain corresponding to the deterministic user behavior demand characterization data (exemplarily, the behavior data description vectors corresponding to the optimized and used exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data can be mean-superimposed to form the distribution domain center data description vector of the corresponding description vector distribution domain);

[0105] Analyze the vector difference between the behavior data description vector corresponding to each of the central exemplary user behavior monitoring image sequences and the central data description vector of the distribution domain (exemplarily, the vector difference may have a negative correlation with the corresponding vector correlation coefficient);

[0106] Based on the vector difference corresponding to each of the central exemplary user behavior monitoring image sequences, in the central behavior data description vector corresponding to the deterministic user behavior demand characterization data, the central behavior data description vector corresponding to the number of discard indicators is discarded (exemplarily, the central behavior data description vector corresponding to the number of discard indicators with the largest vector difference can be discarded).

[0107] Among them, it can be selected that, in some feasible implementations, the low-precision user behavior demand characterization data includes low-precision possibility evaluation parameters corresponding to the deterministic user behavior demand characterization data (the low-precision possibility evaluation parameters, i.e., the possibility evaluation parameters analyzed by the second user behavior analysis neural network that has not undergone network optimization processing, are used to reflect the possibility that the exemplary user behavior monitoring image sequence belongs to the corresponding deterministic user behavior demand characterization data). Based on this, before the step of performing network optimization processing on the second user behavior analysis neural network based on the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form the corresponding optimized user behavior analysis neural network, the network optimization processing of the optimized user behavior analysis neural network may further include the following detailed contents:

[0108] Utilizing the first user behavior analysis neural network, a user behavior analysis operation is performed according to the high-precision behavior data description vector corresponding to each of the exemplary user behavior monitoring image sequences, so as to output high-precision user behavior demand characterization data corresponding to each of the exemplary user behavior monitoring image sequences, wherein the high-precision user behavior demand characterization data includes high-precision possibility assessment parameters corresponding to the deterministic user behavior demand characterization data (as explained above for the low-precision possibility assessment parameters); for each of the exemplary user behavior monitoring image sequences, the low-precision possibility assessment parameters and the corresponding high-precision possibility assessment parameters corresponding to the exemplary user behavior monitoring image sequence are mapped based on the same mapping parameters and mapping rules. Processing is performed to output the mapped low-precision possibility evaluation parameters corresponding to the low-precision possibility evaluation parameters and the mapped high-precision possibility evaluation parameters corresponding to the high-precision possibility evaluation parameters (for example, for each of the exemplary user behavior monitoring image sequences, the ratio between the low-precision possibility evaluation parameters corresponding to the exemplary user behavior monitoring image sequence and the preset parameters can be calculated, and then the ratio can be subjected to an exponential operation to obtain a corresponding exponential operation result, and then the exponential operation result corresponding to the exemplary user behavior monitoring image sequence can be subjected to a ratio processing with the sum of the exponential operation results corresponding to all the exemplary user behavior monitoring image sequences to obtain a mapped low-precision possibility evaluation parameter corresponding to the exemplary user behavior monitoring image sequence). degree possibility assessment parameter, wherein the method for determining the mapping high-precision possibility assessment parameter is consistent with the method for determining the mapping low-precision possibility assessment parameter, and the specific value of the preset parameter is not limited, as long as the method for determining the mapping high-precision possibility assessment parameter is consistent with the method for determining the mapping low-precision possibility assessment parameter); error index analysis is performed based on the mapping low-precision possibility assessment parameter and the mapping high-precision possibility assessment parameter to output a mapping processing error index corresponding to the exemplary user behavior monitoring image sequence (exemplarily, the mapping high-precision possibility assessment parameter can be logarithmically operated, and then the result of the logarithmic operation can be multiplied by the mapping low-precision possibility assessment parameter, and then, The negative correlation value of the multiplication result is used as a mapping processing error index); according to the mapping processing error index corresponding to each of the exemplary user behavior monitoring image sequences, a corresponding cumulative mapping processing error index is determined (based on this, the step of performing network optimization processing on the second user behavior analysis neural network according to the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form a corresponding optimized user behavior analysis neural network may include: performing fusion processing (such as weighted sum calculation, etc.) on the data internal mining error index, the data inter-mining error index, the cumulative mapping processing error index and the behavior demand analysis error index to output a corresponding fusion error index;And, according to the fusion error index, performing network optimization processing on the second user behavior analysis neural network to form a corresponding optimized user behavior analysis neural network);

[0109] Combination Figure 3 The embodiment of the present invention further provides a data processing device based on smart buildings, which can be applied to the above-mentioned data processing system based on smart buildings. The data processing device based on smart buildings (software function device) can include the following functional modules:

[0110] A user behavior monitoring image acquisition module is used to acquire a sequence of user behavior monitoring images to be processed corresponding to a target smart building, wherein the sequence of user behavior monitoring images to be processed includes multiple frames of user behavior monitoring images to be processed, and the multiple frames of user behavior monitoring images to be processed are formed based on image acquisition of at least one monitored building area of ​​the target smart building;

[0111] A user behavior analysis module, used for loading the sequence of user behavior monitoring images to be processed so as to load it into an optimized user behavior analysis neural network, and performing a user behavior analysis operation on the sequence of user behavior monitoring images to be processed by using the optimized user behavior analysis neural network, wherein the optimized user behavior analysis neural network performs network optimization processing on a second user behavior analysis neural network according to multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis basis includes output data of the second user behavior analysis neural network and output data of the first user behavior analysis neural network, wherein the first user behavior analysis neural network is a user behavior analysis neural network that has been subjected to network optimization processing, and the second user behavior analysis neural network is a user behavior analysis neural network that has not been subjected to network optimization processing;

[0112] The user behavior demand output module is used to output the user behavior demand representation data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network, and the user behavior demand representation data is used to reflect the demand information of the building users in the target smart building.

[0113] In summary, the data processing method and system based on smart buildings provided by the present invention can collect the to-be-processed user behavior monitoring image sequence corresponding to the target smart building; use the optimized user behavior analysis neural network to perform user behavior analysis operations on the to-be-processed user behavior monitoring image sequence; and output the user behavior demand characterization data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network. Based on this, since the optimized user behavior analysis neural network performs network optimization processing on the second user behavior analysis neural network based on multi-dimensional error indicators, and in the process of analyzing the error indicators, the analysis basis includes the output data of the second user behavior analysis neural network and the output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to the user behavior analysis neural network that has been processed by network optimization. In this way, the optimized user behavior analysis neural network can not only learn based on the second user behavior analysis neural network, but also learn the information learned by the first user behavior analysis neural network in the process of network optimization processing, so that the analysis reliability of the optimized user behavior analysis neural network is higher, thereby improving the problem of poor reliability of user behavior data analysis in the prior art.

[0114] 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 data processing method based on smart buildings, characterized in that: include: A sequence of user behavior monitoring images to be processed corresponding to a target smart building is collected, wherein the sequence of user behavior monitoring images to be processed includes multiple frames of user behavior monitoring images to be processed, and the multiple frames of user behavior monitoring images to be processed are formed based on image collection of at least one monitored building area of ​​the target smart building; The sequence of user behavior monitoring images to be processed is loaded to be loaded into the optimized user behavior analysis neural network, and the optimized user behavior analysis neural network is used to perform a user behavior analysis operation on the sequence of user behavior monitoring images to be processed, wherein the optimized user behavior analysis neural network performs network optimization processing on a second user behavior analysis neural network according to a multi-dimensional error index, and in the process of analyzing the error index, the analysis basis includes output data of the second user behavior analysis neural network and output data of the first user behavior analysis neural network, the first user behavior analysis neural network belongs to a user behavior analysis neural network that has been subjected to network optimization processing, and the second user behavior analysis neural network belongs to a user behavior analysis neural network that has not been subjected to network optimization processing; Outputting user behavior demand representation data corresponding to the to-be-processed user behavior monitoring image sequence analyzed by the optimized user behavior analysis neural network, wherein the user behavior demand representation data is used to reflect demand information of building users in the target smart building; The network optimization process of optimizing the user behavior analysis neural network includes: An exemplary image sequence cluster is extracted, the exemplary image sequence cluster includes an exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data corresponding to the first user behavior analysis neural network, the exemplary image sequence cluster also includes an exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data, the exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data is used for network optimization processing to form the first user behavior analysis neural network, and the exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data has not been used for network optimization processing to form the first user behavior analysis neural network; Using the first user behavior analysis neural network, mining a high-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster; Using a second user behavior analysis neural network, mining a low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence in the exemplary image sequence cluster, and performing a user behavior analysis operation based on each of the low-precision behavior data description vectors, and then calculating and outputting a corresponding behavior demand analysis error index based on each of the output low-precision user behavior demand representation data; Performing error index analysis based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output a corresponding data internal mining error index; Perform error index analysis based on low-precision behavior data description vectors corresponding to each two of the exemplary user behavior monitoring image sequences having different behavior demand representation data, so as to output corresponding data-to-data mining error index; According to the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index, performing network optimization processing on the second user behavior analysis neural network to form a corresponding optimized user behavior analysis neural network; The step of performing error index analysis based on the low-precision behavior data description vectors corresponding to each two of the exemplary user behavior monitoring image sequences having different behavior demand representation data to output the corresponding data mining error index comprises: Combining each of the exemplary user behavior monitoring image sequences and each of the exemplary user behavior monitoring image sequences in the exemplary image sequence cluster to form a corresponding exemplary user behavior monitoring image sequence combination, so as to form a corresponding plurality of exemplary user behavior monitoring image sequence combinations; Determine the vector correlation coefficient according to the low-precision behavior data description vector corresponding to each of the exemplary user behavior monitoring image sequence combinations; Analyzing the local data mining error index corresponding to each of the exemplary user behavior monitoring image sequence combinations according to the determined vector correlation coefficient; fusing local data-to-data mining error indicators corresponding to each first exemplary user behavior monitoring image sequence combination to output a corresponding data-to-data mining error indicator, wherein the first exemplary user behavior monitoring image sequence combination includes two exemplary user behavior monitoring image sequences having different behavior demand representation data; Before the step of performing network optimization processing on the second user behavior analysis neural network according to the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form a corresponding optimized user behavior analysis neural network, the network optimization processing of the optimized user behavior analysis neural network further includes: Perform error index analysis based on the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to each two exemplary user behavior monitoring image sequences with different behavior demand representation data, so as to output the corresponding inter-data hybrid mining error index; The step of performing network optimization processing on the second user behavior analysis neural network according to the internal data mining error index, the inter-data mining error index and the behavior demand analysis error index to form a corresponding step of optimizing the user behavior analysis neural network includes: The internal data mining error index, the inter-data mining error index, the inter-data mixed mining error index, and the behavior demand analysis error index are fused to output a corresponding fused error index; and, based on the fused error index, the second user behavior analysis neural network is optimized to form a corresponding optimized user behavior analysis neural network; The step of performing error index analysis on the high-precision behavior data description vector and the low-precision behavior data description vector corresponding to the same exemplary user behavior monitoring image sequence to output the corresponding data internal mining error index comprises: Determine a vector correlation coefficient according to the high-precision behavior data description vector and the corresponding low-precision behavior data description vector corresponding to each exemplary user behavior monitoring image sequence, so as to output the corresponding vector correlation coefficient; Analyzing the local data mining error index corresponding to each exemplary user behavior monitoring image sequence according to the determined vector correlation coefficient, wherein there is a non-positive object relationship between the local data mining error index corresponding to the exemplary user behavior monitoring image sequence and the vector correlation coefficient corresponding to the exemplary user behavior monitoring image sequence; Fusing the local data inter-mining error indicators corresponding to each exemplary user behavior monitoring image sequence to output the corresponding data internal mining error indicators; The deterministic user behavior demand characterization data includes a variety of different user behavior demand characterization data, and the step of extracting an exemplary image sequence cluster includes: Extracting an original exemplary image sequence cluster, the original exemplary image sequence cluster includes a central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data, and the projection coordinates of the central exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data in the description vector distribution domain corresponding to the deterministic user behavior demand characterization data are associated with the projection coordinates of all exemplary image sequences corresponding to the deterministic user behavior demand characterization data in the description vector distribution domain; Extracting an extended exemplary user behavior monitoring image sequence cluster corresponding to the extended user behavior demand characterization data, and forming an exemplary image sequence cluster based on the extended exemplary user behavior monitoring image sequence cluster and the original exemplary image sequence cluster; Before the step of extracting the original exemplary image sequence cluster, the network optimization process of optimizing the user behavior analysis neural network further includes: Extracting all exemplary image sequences corresponding to each deterministic user behavior demand representation data, wherein all exemplary image sequences include an optimized exemplary user behavior monitoring image sequence corresponding to the first user behavior analysis neural network; For all exemplary image sequences corresponding to each of the deterministic user behavior demand characterization data, classify the behavior data description vectors of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand characterization data; In each of the data classification sets, a behavior data description vector that matches the corresponding classification center behavior data description vector is determined, and the optimized exemplary user behavior monitoring image sequence corresponding to the determined behavior data description vector is marked as the center exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data; The step of classifying the behavior data description vector of each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences to form a plurality of data classification sets corresponding to the deterministic user behavior demand characterization data includes: Extracting the number of classification indicators corresponding to the deterministic user behavior demand representation data; Classify the behavior data description vector corresponding to each optimized exemplary user behavior monitoring image sequence in all the exemplary image sequences according to the number of classification indicators to form a plurality of data classification sets whose number of corresponding deterministic user behavior demand representation data is consistent with the number of classification indicators; The amount of the extracted extensible user behavior demand characterization data is greater than 1, and after forming the optimized user behavior analysis neural network, the method further includes: Counting the number of demand types corresponding to the extended user behavior demand representation data, and discarding a plurality of central exemplary user behavior monitoring image sequences in the original exemplary image sequence cluster according to the number of demand types; Analyzing a central exemplary user behavior monitoring image sequence corresponding to the extended user behavior demand characterization data from the extended exemplary user behavior monitoring image sequence cluster corresponding to the extended user behavior demand characterization data; Saving a central exemplary user behavior monitoring image sequence corresponding to the extensible user behavior demand characterization data; The step of discarding a plurality of central exemplary user behavior monitoring image sequences in the original exemplary image sequence cluster according to the number of demand types includes: According to the number of demand types, the number of discarding indicators of the exemplary user behavior monitoring image sequence to be discarded corresponding to each deterministic user behavior demand representation data is analyzed, wherein the number of demand types is used as the corresponding number of discarding indicators, or a positive correlation value of the number of demand types is used as the corresponding number of discarding indicators; For each of the deterministic user behavior demand characterization data, according to the optimized and used exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data, a distribution domain center data description vector of the description vector distribution domain corresponding to the deterministic user behavior demand characterization data is analyzed, wherein the behavior data description vectors corresponding to the optimized and used exemplary user behavior monitoring image sequence corresponding to the deterministic user behavior demand characterization data are mean-superimposed to form a distribution domain center data description vector of the corresponding description vector distribution domain; Analyze the vector difference between the behavior data description vector corresponding to each of the central exemplary user behavior monitoring image sequences and the central data description vector of the distribution domain, wherein the vector difference has a negative correlation with the corresponding vector correlation coefficient; According to the vector difference corresponding to each of the central exemplary user behavior monitoring image sequences, in the central behavior data description vector corresponding to the deterministic user behavior demand representation data, the central behavior data description vector corresponding to the discarding indicator number is discarded; Wherein, the low-precision user behavior demand characterization data includes low-precision possibility evaluation parameters corresponding to the deterministic user behavior demand characterization data, and the low-precision possibility evaluation parameters are possibility evaluation parameters analyzed by using the second user behavior analysis neural network that has not undergone network optimization processing, and are used to reflect the possibility that the exemplary user behavior monitoring image sequence belongs to the corresponding deterministic user behavior demand characterization data. Before the step of performing network optimization processing on the second user behavior analysis neural network based on the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form the corresponding optimized user behavior analysis neural network, the network optimization processing of the optimized user behavior analysis neural network also includes: Using the first user behavior analysis neural network, a user behavior analysis operation is performed according to the high-precision behavior data description vector corresponding to each of the exemplary user behavior monitoring image sequences, so as to output high-precision user behavior demand representation data corresponding to each of the exemplary user behavior monitoring image sequences, wherein the high-precision user behavior demand representation data includes high-precision possibility evaluation parameters corresponding to the deterministic user behavior demand representation data; For each of the exemplary user behavior monitoring image sequences, the low-precision possibility assessment parameters and the corresponding high-precision possibility assessment parameters corresponding to the exemplary user behavior monitoring image sequence are mapped based on the same mapping parameters and mapping rules to output the mapped low-precision possibility assessment parameters corresponding to the low-precision possibility assessment parameters and the mapped high-precision possibility assessment parameters corresponding to the high-precision possibility assessment parameters, wherein for each of the exemplary user behavior monitoring image sequences, the ratio between the low-precision possibility assessment parameters corresponding to the exemplary user behavior monitoring image sequence and the preset parameters is calculated, and then the ratio is exponentially operated to obtain the corresponding exponential operation result, and then the exponential operation result corresponding to the exemplary user behavior monitoring image sequence is ratioed with the sum of the exponential operation results corresponding to all the exemplary user behavior monitoring image sequences to obtain the mapped low-precision possibility assessment parameters corresponding to the exemplary user behavior monitoring image sequence, wherein the method for determining the mapped high-precision possibility assessment parameters is consistent with the method for determining the mapped low-precision possibility assessment parameters; performing error index analysis according to the mapping low-precision possibility evaluation parameter and the mapping high-precision possibility evaluation parameter to output a mapping processing error index corresponding to the exemplary user behavior monitoring image sequence, wherein specifically performing a logarithmic operation on the mapping high-precision possibility evaluation parameter, and then multiplying the result of the logarithmic operation by the mapping low-precision possibility evaluation parameter, and then taking the negative correlation value of the multiplication result as the mapping processing error index; Determining a corresponding cumulative mapping processing error index according to the mapping processing error index corresponding to each of the exemplary user behavior monitoring image sequences; The step of performing network optimization processing on the second user behavior analysis neural network according to the data internal mining error index, the data inter-mining error index and the behavior demand analysis error index to form a corresponding optimized user behavior analysis neural network includes: The internal data mining error index, the inter-data mining error index, the cumulative mapping processing error index and the behavior demand analysis error index are fused to output a corresponding fusion error index; and, based on the fusion error index, the second user behavior analysis neural network is network optimized to form a corresponding optimized user behavior analysis neural network.

2. A data processing system based on smart buildings, characterized in that: It includes a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to implement the data processing method based on smart buildings as described in claim 1.

Images