Energy station-oriented multi-source heterogeneous data fusion twinborn model construction method

Abstract

The invention discloses an energy station-oriented multi-source heterogeneous data fusion twinborn model construction method, which belongs to the technical field of model construction and comprises the following steps of S1, acquiring static physical space parameterized model data of an energy station; s2, acquiring non-parametric model data of the dynamic energy flow of the energy station; s3, extracting the characteristics of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parameterized model data of the energy station through a multi-dimensional scale transformation method; s4, multi-source heterogeneous data fusion expression; s5, constructing a digital twinborn fusion model, training the digital twinborn fusion model through the data of the multi-source heterogeneous data fusion expression until convergence, and obtaining the digital twinborn fusion model; according to the method, multi-source heterogeneous data can be effectively fused to accurately perceive and analyze the operation situation of the energy station, so that the geometrical characteristics, physical characteristics and energy operation rules of the energy station can be comprehensively described to support mapping and interaction between a virtual space and a real space.

Description

Technical Field

[0001] The present invention belongs to the technical field of model construction, and specifically relates to a method for constructing a multi-source heterogeneous data fusion twin model for an energy station. Background Art

[0002] With the rapid development of energy Internet, energy stations, as important nodes for coupling multiple energy sources, have a direct impact on the stability and security of the entire energy system due to their operating status and efficiency.

[0003] However, in actual operation, energy stations face challenges such as diverse data sources, complex data structures, and different time scales. The lack of effective correlation and fusion mechanisms between multi-source data makes it difficult to accurately perceive and analyze the operating status of energy stations.

[0004] In view of this, a method for constructing a multi-source heterogeneous data fusion twin model for energy stations is designed to solve the above problems. Summary of the invention

[0005] To solve the problems raised in the above background technology. The present invention provides a method for constructing a multi-source heterogeneous data fusion twin model for an energy station, which can effectively fuse multi-source heterogeneous data to accurately perceive and analyze the operation status of the energy station, thereby being able to achieve a comprehensive characterization of the geometric characteristics, physical properties and energy operation laws of the energy station, so as to support the mapping and interaction of virtual space and real space.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for constructing a multi-source heterogeneous data fusion twin model for an energy station, comprising the following steps:

[0007] S1: Establish a system logic model through TRNSYS software, establish an energy station BIM model through REVIT software, convert and analyze model data formats through IFC standard, and obtain the static physical space parametric model data of the energy station;

[0008] S2: Deploy the IoT management system through the energy station, deploy IoT devices and IoT edge computing nodes at key locations of the energy station, and the IoT devices and IoT edge computing nodes transmit the original operation data and sensor node metadata to the IoT management system through the wireless network. The IoT management system parses and obtains the non-parametric model data of the dynamic energy flow of the energy station;

[0009] S3: The features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station are extracted by multidimensional scaling transformation method. The features of the static physical space parameterized model data of the energy station include geometric features and physical features. The geometric features include the shape, size and position of the equipment. The physical features include material properties, thermal conductivity and density. The features of the dynamic energy flow non-parametric model data of the energy station include time series features and frequency domain features. The time series features include temperature, pressure, flow, humidity and electric energy data collected by sensors.

[0010] S4: Based on the extracted features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station, multi-source heterogeneous data fusion expression is performed;

[0011] S5: Build a digital twin fusion model, train the digital twin fusion model through the data expressed by multi-source heterogeneous data fusion until convergence, and obtain the digital twin fusion model.

[0012] Furthermore, the specific steps of step S4 include:

[0013] S41: Normalize the extracted static physical space parametric model data of the energy station, i.e., the features of the BIM model data

[0014] Assuming that the geometric features of the original BIM model data are (x, y, z), the normalization process is completed by the following formula:

[0015]

[0016] Where: (x min ,y min ,z min ) and (x max ,y max ,z max ) represent the minimum and maximum values ​​of the geometric features respectively;

[0017] S42: Normalize the extracted non-parametric model data of the dynamic energy flow of the energy station, that is, the characteristics of the IOT device data

[0018] Assuming that the time series feature of IoT device data is a time series d(t), the normalization process is completed by the following formula:

[0019]

[0020] Where: d min and d max are the minimum and maximum values ​​of the sensor data, respectively;

[0021] S43: Spatial registration of the normalized BIM model and the features of IoT device data

[0022] Assume that the center coordinates of the BIM model are (x BIM ,y BIM ,z BIM ), the center coordinate of the IOT device is (x IOT ,y IOT ,z IOT ), the transformation matrix is ​​T, and it is aligned to the same space through the space vector method. The spatial registration formula is:

[0023]

[0024] S44: Time alignment of the features of the normalized BIM model and IoT device data

[0025] Assume that there are two dimensions of time series data, namely BIM data B(t) and IOT data I(t). The two dimensions of data are aligned to the same time point t by interpolation. The time alignment formula is:

[0026]

[0027] Where: Δt is the time step, t start is the starting time;

[0028] S45: Multimodal Data Fusion

[0029] The multimodal data is fused by weighted average method, and the formula is:

[0030] F=w g ·G+w p ·P+w i I

[0031] Where: G is the normalized geometric data, P is the physical attribute data, I is the IOT sensor data, and w g 、w p and w i is the corresponding preset adjustable weight coefficient.

[0032] 3. According to the method for constructing a multi-source heterogeneous data fusion twin model for an energy station according to claim 1, it is characterized in that: in the step S43, the calculation formula of the center coordinate is:

[0033] Assume that the geometric features of an object are (x, y, z), and its geometric center, that is, the center coordinates (x c ,y c ,z c )for:

[0034]

[0035] Where: x i ,y i and z i It is represented by the current geometric features, namely x, y and z, which are obtained by calculating the arithmetic average of the corresponding coordinates of all points in the object, and N is the total number of points.

[0036] 4. The method for constructing a multi-source heterogeneous data fusion twin model for an energy station according to claim 1, characterized in that: in step S5, the formula of the digital twin fusion model is:

[0037]

[0038] Where: It is represented as the prediction result, f is represented as the conversion parameter from input to output, F is represented as the fusion feature, and θ is represented as the model parameter.

[0039] 5. The method for constructing a multi-source heterogeneous data fusion twin model for an energy station according to claim 1, characterized in that: in step S5, the optimization formula of the model parameter θ of the digital twin fusion model is:

[0040] θ t+1 =θ t -η▽ θ L(θ t )

[0041] Where: θ t+1 Represented as the model parameter vector at iteration t+1, θ t It is represented as the model parameter vector at the tth iteration, η is the learning efficiency, which determines the step size of each model update.

[0042] ▽ θ L(θ t ) is expressed as the loss function L(θ t ) for the parameter θ t The gradient of L(θ t ) is expressed as the loss function, and is expressed as the current parameter θ t The error of the model below.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention first constructs a system logic model and an energy station BIM model, transforms and parses model data based on the IFC standard to obtain static physical space parameterized model data of the energy station, then constructs an Internet of Things management system, an Internet of Things device and an Internet of Things edge computing node, and parses the data transmitted by the Internet of Things device and the Internet of Things edge computing node based on the Internet of Things management system to obtain dynamic energy flow non-parametric model data of the energy station, then extracts target features based on the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station, then performs multi-source heterogeneous data fusion expression based on the extracted target features, and then constructs a digital twin fusion model, and trains the digital twin fusion model with data expressed by multi-source heterogeneous data fusion until convergence to obtain a digital twin fusion model. The digital twin fusion model constructed by the construction method can effectively fuse multi-source heterogeneous data to accurately perceive and analyze the operation status of the energy station, thereby being able to achieve a comprehensive characterization of the geometric features, physical characteristics and energy operation laws of the energy station, so as to support the mapping and interaction between virtual space and real space. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of the method for constructing a multi-source heterogeneous data fusion twin model for energy stations according to the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1 The present invention provides the following technical solution: a method for constructing a multi-source heterogeneous data fusion twin model for an energy station, comprising the following steps:

[0048] S1: Establish a system logic model through TRNSYS software, establish an energy station BIM model through REVIT software, convert and analyze model data formats through IFC standard, and obtain the static physical space parametric model data of the energy station;

[0049] S2: Deploy the IoT management system through the energy station, deploy IoT devices and IoT edge computing nodes at key locations of the energy station, and the IoT devices and IoT edge computing nodes transmit the original operation data and sensor node metadata to the IoT management system through the wireless network. The IoT management system parses and obtains the non-parametric model data of the dynamic energy flow of the energy station;

[0050] S3: The features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station are extracted by multidimensional scaling transformation method. The features of the static physical space parameterized model data of the energy station include geometric features and physical features. The geometric features include the shape, size and position of the equipment. The physical features include material properties, thermal conductivity and density. The features of the dynamic energy flow non-parametric model data of the energy station include time series features and frequency domain features. The time series features include temperature, pressure, flow, humidity and electric energy data collected by sensors.

[0051] Assuming that the geometric features of an object are (x, y, z), the geometric center (x c ,y c ,z c ) and volume V and other characteristics;

[0052] Geometric center (x c ,y c ,z c ) The formula of the characteristic is shown below;

[0053] The formula for the volume V characteristic is:

[0054]

[0055] Assuming that the physical characteristics of an object to be extracted are (m, k, ρ), m is the material, k is the thermal conductivity, and ρ is the density. The thermal conductivity equation q and mass M can be calculated by expanding the physical characteristics (m, k, ρ);

[0056] The heat conduction equation q is:

[0057] q=-k▽T;

[0058] The mass M is:

[0059] M = ρ·V;

[0060] Assuming that the time series data of an object is d(t), statistical features such as mean μ and variance σ can be calculated by expanding the time series data d(t). 2 wait;

[0061] Mean μ:

[0062]

[0063] Variance σ 2 :

[0064]

[0065] The time series data is converted to the frequency domain through Fourier transform to extract the frequency domain features. Assuming that the time series data is d(t), its Fourier transform is:

[0066]

[0067] The energy spectrum is:

[0068] P(f)=|D(f)| 2 ;

[0069] S4: Based on the extracted features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station, multi-source heterogeneous data fusion expression is performed;

[0070] S41: Normalize the extracted static physical space parametric model data of the energy station, i.e., the features of the BIM model data

[0071] Assuming that the geometric features of the original BIM model data are (x, y, z), the normalization process is completed by the following formula:

[0072]

[0073] Where: (x min ,y min ,z min ) and (x max ,y max ,z max ) represent the minimum and maximum values ​​of the geometric features respectively;

[0074] S42: Normalize the extracted non-parametric model data of the dynamic energy flow of the energy station, that is, the characteristics of the IOT device data

[0075] Assuming that the time series feature of IoT device data is a time series d(t), the normalization process is completed by the following formula:

[0076]

[0077] Where: d min and d max are the minimum and maximum values ​​of the sensor data, respectively;

[0078] S43: Spatial registration of the normalized BIM model and the features of IoT device data

[0079] The calculation formula of the center coordinates is:

[0080] Assume that the geometric features of an object are (x, y, z), and its geometric center, that is, the center coordinates (x c ,y c ,zc )for:

[0081]

[0082] Where: x i ,y i and z i It is represented by the current geometric features, i.e. x, y and z, which are obtained by calculating the arithmetic average of the corresponding coordinates of all points in the object, and N is the total number of points;

[0083] Assume that the center coordinates of the BIM model are (x BIM ,y BIM ,z BIM ), the center coordinate of the IOT device is (x IOT ,y IOT ,z IOT ), the transformation matrix is ​​T, and it is aligned to the same space through the space vector method. The spatial registration formula is:

[0084]

[0085] S44: Time alignment of the features of the normalized BIM model and IoT device data

[0086] Assume that there are two dimensions of time series data, namely BIM data B(t) and IOT data I(t). The two dimensions of data are aligned to the same time point t by interpolation. The time alignment formula is:

[0087]

[0088] Where: Δt is the time step, t start is the starting time;

[0089] S45: Multimodal Data Fusion

[0090] The multimodal data is fused by weighted average method, and the formula is:

[0091] F=w g ·G+w p ·P+w i I

[0092] Where: G is the normalized geometric data, P is the physical attribute data, I is the IOT sensor data, and w g 、w p and w i is the corresponding preset adjustable weight coefficient;

[0093] S5: Build a digital twin fusion model, train the digital twin fusion model through the data expressed by multi-source heterogeneous data fusion until convergence, and obtain the digital twin fusion model;

[0094] The formula of the digital twin fusion model is:

[0095]

[0096] Where: It is represented as the prediction result, f is represented as the conversion parameter from input to output, F is represented as the fusion feature, and θ is represented as the model parameter;

[0097] The optimization formula of the model parameter θ is:

[0098] θ t+1 =θ t -η▽ θ L(θ t )

[0099] Where: θ t+1 Represented as the model parameter vector at iteration t+1, θ t It is represented as the model parameter vector at the tth iteration, η is the learning efficiency, which determines the step size of each model update.

[0100] ▽ θ L(θ t ) is expressed as the loss function L(θ t ) for the parameter θ t The gradient of L(θ t ) is expressed as the loss function, and is expressed as the current parameter θ t The error of the model below.

[0101] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for constructing a multi-source heterogeneous data fusion twin model for an energy station, characterized in that: The following steps are involved: S1: Establish a system logic model through TRNSYS software, establish an energy station BIM model through REVIT software, convert and analyze model data formats through IFC standard, and obtain static physical space parametric model data of the energy station; S2: Deploy the IoT management system through the energy station, deploy IoT devices and IoT edge computing nodes at key locations of the energy station, and the IoT devices and IoT edge computing nodes transmit the original operation data and sensor node metadata to the IoT management system through the wireless network. The IoT management system parses and obtains the non-parametric model data of the dynamic energy flow of the energy station; S3: The features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station are extracted by multidimensional scaling transformation method. The features of the static physical space parameterized model data of the energy station include geometric features and physical features. The geometric features include the shape, size and position of the equipment. The physical features include material properties, thermal conductivity and density. The features of the dynamic energy flow non-parametric model data of the energy station include time series features and frequency domain features. The time series features include temperature, pressure, flow, humidity and electric energy data collected by sensors. S4: Based on the extracted features of the static physical space parameterized model data of the energy station and the dynamic energy flow non-parametric model data of the energy station, multi-source heterogeneous data fusion expression is performed; S5: Build a digital twin fusion model, train the digital twin fusion model through the data expressed by multi-source heterogeneous data fusion until convergence, and obtain the digital twin fusion model.

2. According to claim 1, a method for constructing a multi-source heterogeneous data fusion twin model for an energy station is characterized by: The specific steps of step S4 include: S41: Normalize the extracted static physical space parametric model data of the energy station, i.e., the features of the BIM model data Assuming that the geometric features of the original BIM model data are (x, y, z), the normalization process is completed by the following formula: Where: (x min ,y min ,z min ) and (x max ,y max ,z max ) represent the minimum and maximum values ​​of the geometric features respectively; S42: Normalize the extracted non-parametric model data of the dynamic energy flow of the energy station, that is, the characteristics of the IOT device data Assuming that the time series feature of IoT device data is a time series d(t), the normalization process is completed by the following formula: Where: d min and d max are the minimum and maximum values ​​of the sensor data, respectively; S43: Spatial registration of the normalized BIM model and the features of IoT device data Assume that the center coordinates of the BIM model are (x BIM ,y BIM ,z BIM ), the center coordinate of the IOT device is (x IOT ,y IOT ,z IOT ), the transformation matrix is ​​T, and it is aligned to the same space through the space vector method. The spatial registration formula is: S44: Time alignment of the features of the normalized BIM model and IoT device data Assume that there are two dimensions of time series data, namely BIM data B(t) and IOT data I(t). The two dimensions of data are aligned to the same time point t by interpolation. The time alignment formula is: Where: Δt is the time step, t start is the starting time; S45: Multimodal Data Fusion The multimodal data is fused by weighted average method, and the formula is: F=w g ·G+w p ·P+w i ·I Where: G is the normalized geometric data, P is the physical attribute data, I is the IOT sensor data, and w g 、w p and w i is the corresponding preset adjustable weight coefficient.

3. According to claim 1, a method for constructing a multi-source heterogeneous data fusion twin model for an energy station is characterized in that: In step S43, the calculation formula of the center coordinates is: Assume that the geometric features of an object are (x, y, z), and its geometric center, that is, the center coordinates (x c ,y c ,z c )for: Where: x i ,y i and z i It is represented by the current geometric features, namely x, y and z, which are obtained by calculating the arithmetic average of the corresponding coordinates of all points in the object, and N is the total number of points.

4. The method for constructing a multi-source heterogeneous data fusion twin model for an energy station according to claim 1, characterized in that: In step S5, the formula of the digital twin fusion model is: Where: It is represented as the prediction result, f is represented as the conversion parameter from input to output, F is represented as the fusion feature, and θ is represented as the model parameter.

5. The method for constructing a multi-source heterogeneous data fusion twin model for an energy station according to claim 1, characterized in that: In step S5, the optimization formula of the model parameter θ of the digital twin fusion model is: i t+1 =θ t -η▽ θ L(θ t ) Where: θ t+1 Represented as the model parameter vector at iteration t+1, θ t is represented as the model parameter vector at the tth iteration, η is the learning efficiency, which determines the step size of each model update, ▽ θ L(θ t ) is expressed as the loss function L(θ t ) for the parameter θ t The gradient of L(θ t ) is expressed as the loss function, and is expressed as the current parameter θ t The error of the model below.

Images