Portable transformer area situation awareness analysis device

Abstract

The invention relates to a portable transformer area situation awareness analysis device which comprises a box body, a power supply misconnection prevention protection device is installed in the middle of the upper surface of the box body, a power supply is installed above the box body in a sliding mode, an installation frame is arranged in front of the power supply, and a sensor module is installed on one side of the upper surface of the installation frame. A universal adjusting illuminating lamp is arranged on the top surface of the sensor module; a man-machine interaction mechanism is installed on one side edge of the installation frame, a two-way communication module is arranged behind the man-machine interaction mechanism, and the two-way communication module is in sliding connection with the installation frame and can stretch out and draw back into the installation frame; a data acquisition and processing module is arranged on one side of the front of the mounting frame, and a standby UPS is arranged on the right side of the data acquisition and processing module; the system has the advantages of being simple and reasonable in structure, convenient and fast to use, and capable of achieving transformer area real-time state estimation, line situation awareness evaluation and transformer area maintenance strategy optimization.

Description

Technical Field

[0001] The present invention belongs to the technical field of electrical equipment, and particularly relates to a portable analysis device for substation area situation awareness. Background Art

[0002] With the rapid development of economy and society, the architecture and operation of low-voltage substations have become increasingly complex. When grid personnel conduct inspection and maintenance work on line loss, electricity theft, leakage, etc., the workload is extremely large. In order to more effectively monitor substations, reduce the burden on grid personnel, and ensure power supply reliability, economy, and security, currently, portable low-voltage substation monitoring devices are used, which can perform real-time online detection, local storage, and long-distance wireless transmission of the current of multiple branch lines covered by low-voltage substations, saving more time costs, labor costs, and funding costs. However, with the continuous increase in the penetration rate of distributed renewable energy, the uncertainty and volatility of transformer power supply substations have also been increasing. In this case, due to the lack of supporting detection devices in existing portable low-voltage substation monitoring devices, it is not convenient to sense the topology of the power grid, resulting in the inability to accurately and real-time monitor the system state of the low-voltage distribution substation. In addition, since the low-voltage substation is at the end of the distribution network and at the lowest level of the power system, compared with the high-voltage power grid, the influence range of the substation is relatively small. Therefore, the investment and attention on the operation and maintenance of the substation are relatively weak. Existing portable low-voltage substation monitoring devices have always followed the methods of planned maintenance and emergency repair after the event for the operation and maintenance and repair work of substations. The decision-making mechanism is relatively simple, and the purpose and pertinence of the repair are not clear enough. In addition, the human resources for the distribution network operation and maintenance of power supply companies are limited, and the effect of planned maintenance is not good, resulting in frequent emergency repair events after the event and causing load losses. Therefore, it is very necessary to provide a portable analysis device for substation area situation awareness with a simple and reasonable structure, convenient and fast use, realizing real-time state estimation of substations, line situation awareness evaluation, and optimizing the substation repair strategy. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a portable analysis device for substation area situation awareness with a simple and reasonable structure, convenient and fast use, realizing real-time state estimation of substations, line situation awareness evaluation, and optimizing the substation repair strategy.

[0004] The object of the present invention is achieved as follows: A portable power distribution area situation awareness analysis device, which includes a box body, and is characterized in that: A power supply anti-misconnection protection device is installed in the middle of the upper surface of the box body; A power supply is slidably installed above the box body; An installation frame is arranged in front of the power supply; A sensor module is installed on one side of the upper surface of the installation frame; A universal adjustable lighting lamp is arranged on the top surface of the sensor module; A human-computer interaction mechanism is installed on one side edge of the installation frame; A two-way communication module is arranged behind the human-computer interaction mechanism, and the two-way communication module is slidably connected to the installation frame and can be retracted into the installation frame; A data acquisition and processing module is arranged on one side in front of the installation frame, and a standby UPS power supply is arranged on the right side of the data acquisition and processing module.

[0005] Bottom pads are arranged at the four corners of the bottom of the box body; An expansion interface board is arranged on the front of the box body, and the expansion interface board includes a temperature and humidity interface, an infrared interface, an environmental monitoring interface, and other standby interfaces; Slide rails are arranged on both side edges of the upper surface of the box body; A main switch is arranged at one corner of the upper surface of the box body; A situation awareness analysis module and a fault diagnosis and warning module are arranged inside the box body.

[0006] The power supply includes a power module, a bottom plate is arranged at the bottom of the power module, and T-shaped sliding plates slidably connected to the corresponding slide rails are arranged on both sides of the lower surface of the bottom plate.

[0007] The sensor module is used to collect power distribution area data, and includes a current transformer, a voltage transformer, a power sensor, an image sensor, and a smoke sensor.

[0008] The human-computer interaction mechanism includes a C-shaped frame arranged on the side edge of the installation frame, a rotating seat is arranged at the free end above the C-shaped frame, a human-computer interaction large screen for providing an operation interface and data display is arranged inside the rotating seat, and the rotating seat is movably connected to one side edge of the human-computer interaction large screen through a universal ball seat device.

[0009] The two-way communication module is used to realize data transmission between the device and an external system, and includes a two-way communicator. An outer frame board is installed outside the two-way communicator, a handle is arranged on the top of the outer frame board, hinge plates are hingedly installed on both side surfaces at the lower ends of the outer frame board, and the hinge plates are slidably connected to a sliding channel located inside the installation frame; Heat dissipation openings are arranged on the outer side surface of the data acquisition and processing module, and heat dissipation fans are installed inside the heat dissipation openings; The data acquisition and processing module is used to process and analyze the data of the sensor module, and includes a real-time data acquisition module, a data normalization processing module, and an algorithm library.

[0010] Advantages of the present invention: The present invention is a portable power distribution network area situation awareness analysis device. During use, the sensor module of the present invention is responsible for collecting power distribution network area environment data; the data acquisition and processing module processes and analyzes the sensor data, and multiple algorithms are built into the algorithm library of the data acquisition and processing module for fault diagnosis and status assessment; the two-way communication module realizes data transmission between the device and an external system; the human-computer interaction mechanism provides an operation interface and data display; the power supply module and the box body are connected by a sliding plate and a slide rail, which is convenient for replacing or regularly maintaining the power supply module, and the standby UPS power supply is used as a backup power supply; the power supply anti-misconnection protection device effectively avoids damage to the device caused by misconnection of 380V. The present invention has the advantages of simple and reasonable structure, convenient and fast use, realizing real-time state estimation of the power distribution network area, line situation awareness assessment, and optimizing the power distribution network area maintenance strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Fig. is the front view of the present invention.

[0012] Figure 2 Fig. is the structural schematic diagram of the box body of the present invention.

[0013] Figure 3 Fig. is the structural schematic diagram of the mounting rack of the present invention Figure 1 .

[0014] Figure 4 Fig. is the structural schematic diagram of the mounting rack of the present invention Figure 2 .

[0015] Figure 5 Fig. is the structural schematic diagram of the mounting rack of the present invention Figure 3 .

[0016] Figure 6 Fig. is the principle block diagram of the present invention.

[0017] Figure 7 Fig. is the composition block diagram of the sensor module of the present invention.

[0018] Figure 8 Fig. is the composition block diagram of the data acquisition and processing module of the present invention.

[0019] Figure 9 Fig. is the composition block diagram of the situation awareness analysis module of the present invention.

[0020] Figure 10 Fig. is the relationship diagram between the data normalization processing module and the real-time data acquisition module of the present invention.

[0021] Figure 11 Fig. is the schematic diagram of the hybrid multiplier operation of STFGCN of the present invention.

[0022] Figure 12Schematic diagram of the distribution network state estimation model based on STFGCN of the present invention.

[0023] Figure 13 Flow chart of the distribution network state estimation model based on STFGCN of the present invention.

[0024] Figure 14 Equivalent circuit diagram of the power supply of the transformer area of the present invention.

[0025] Figure 15 Schematic diagram of the change trend of the impedance of the main and branch lines of the transformer area users of the present invention with the electrical distance from the user to the distribution transformer.

[0026] Figure 16 Relationship diagram between the maintenance investment and the duration of the maintenance trigger under the state maintenance strategy of the present invention.

[0027] In the figure: 1. Box body; 11. Bottom pad; 12. Extended interface board; 13. Main switch; 14. Slide rail; 2. Power supply; 21. Power module; 22. Bottom plate; 23. T-shaped slide plate; 3. Mounting rack; 4. Sensor module; 5. Universal adjustable lighting lamp; 6. Human-computer interaction mechanism; 61. C-shaped frame; 62. Rotating seat; 63. Human-computer interaction large screen; 64. Universal ball seat device; 7. Bidirectional communication module; 71. Bidirectional communicator; 72. Outer frame plate; 73. Handle; 74. Hinge plate; 75. Sliding channel; 8. Data acquisition and processing module; 81. Heat dissipation port; 82. Heat dissipation fan; 9. Backup UPS power supply; 10. Power anti-misconnection protection device. Specific implementation mode

[0028] The present invention will be further described below with reference to the accompanying drawings.

[0029] Embodiment 1

[0030] As Figures 1-16 shown, a portable transformer area situation awareness analysis device includes a box body 1, a power anti-misconnection protection device 10 is installed in the middle of the upper surface of the box body 1, a power supply 2 is slidably installed above the box body 1, an installation rack 3 is arranged in front of the power supply 2, a sensor module 4 is installed on one side of the upper surface of the installation rack 3, and a universal adjustable lighting lamp 5 is arranged on the top surface of the sensor module 4; a human-computer interaction mechanism 6 is installed on one side of the installation rack 3, a bidirectional communication module 7 is arranged behind the human-computer interaction mechanism 6, and the bidirectional communication module 7 is slidably connected to the installation rack 3 and can be retracted into the installation rack 3; a data acquisition and processing module 8 is arranged on one side in front of the installation rack 3, and a backup UPS power supply 9 is arranged on the right side of the data acquisition and processing module 8.

[0031] Four bottom corners of the box body 1 are each provided with a bottom pad 11. An expansion interface board 12 is provided in front of the box body 1. The expansion interface board 12 includes a temperature and humidity interface, an infrared interface, an environmental monitoring interface, and other spare interfaces. Both side edges of the upper surface of the box body 1 are provided with slide rails 14. A main switch 13 is provided at one corner of the upper surface of the box body 1. A situation awareness analysis module and a fault diagnosis and early warning module are provided inside the box body 1.

[0032] In this embodiment, the fault diagnosis and early warning module monitors air switch tripping and power outage and power restoration events, and through analyzing and processing relevant data information of the situation awareness analysis module, combined with the functions of sound and light alarm and voice intercom, realizes rapid fault location and risk early warning, realizes real-time monitoring, data collection and intelligent analysis of the power distribution network (especially the low-voltage substation area), can quickly identify the topology relationship of the substation area, locate faults, analyze line losses and optimize power consumption management.

[0033] The power supply 2 includes a power module 21. A bottom plate 22 is provided at the bottom of the power module 21. T-shaped sliding plates 23 slidably connected to the corresponding slide rails 14 are provided on both sides of the lower surface of the bottom plate 22.

[0034] The sensor module 4 is used to collect substation area data, including current transformers, voltage transformers, power sensors, image sensors, and smoke sensors.

[0035] In this embodiment, the current transformer can adopt a permalloy core current transformer, with full-range automatic gain adjustment to suppress high-frequency harmonic interference.

[0036] The human-computer interaction mechanism 6 includes a C-shaped frame 61 provided on the side of the mounting frame 3. A rotating seat 62 is provided at the free end above the C-shaped frame 61. A human-computer interaction large screen 63 for providing an operation interface and data display is provided inside the rotating seat 62. The rotating seat 62 is movably connected to one side edge of the human-computer interaction large screen 63 through a universal ball seat device 64.

[0037] The two-way communication module 7 is used to realize data transmission between the device and an external system, including a two-way communicator 71. An outer frame plate 72 is installed outside the two-way communicator 71. A handle 73 is provided at the top of the outer frame plate 72. Hinged plates 74 are hinged and installed on both side surfaces at the lower ends of the outer frame plate 72. The hinged plates 74 are slidably connected to a sliding channel 75 inside the mounting frame 3. A heat dissipation port 81 is provided on the outer side surface of the data collection and processing module 8. A heat dissipation fan 82 is installed inside the heat dissipation port 81. The data collection and processing module 8 is used to process and analyze the data of the sensor module 4, including a real-time data collection module, a data normalization processing module, and an algorithm library.

[0038] In this embodiment, the bidirectional communication module supports wired (such as Ethernet) and wireless (such as Wi-Fi, 4G / 5G) communications; and supports multiple protocols, such as Modbus, TCP / IP, etc., to ensure compatibility with external systems;

[0039] The bidirectional communicator adopts bidirectional communication technologies (such as power frequency zero-crossing communication, combination of FSK carrier and pulse current) to ensure accurate identification of the user's affiliated substation area and phase in complex lines (shared high voltage, shared ground, shared cable trench), avoid misjudgment, and the accuracy rate reaches 100%; at the same time, it supports multi-mode identification methods (power line carrier, pulse current, integrated mode), adapts to different on-site requirements, and can simultaneously distinguish up to 12 branch lines.

[0040] The present invention is a portable substation area situation awareness analysis device. During use, the present invention is responsible for collecting substation area environmental data through the sensor module 4; processes and analyzes the sensor data through the data acquisition and processing module 8, and the algorithm library of the data acquisition and processing module 8 has a variety of algorithms for fault diagnosis and status evaluation; realizes data transmission between the device and the external system through the bidirectional communication module 7; provides an operation interface and data display through the human-computer interaction mechanism 6; the power supply module 2 and the box body 1 are connected by a sliding plate and a sliding rail, which is convenient for replacing or regularly maintaining the power supply module 2, and the standby UPS power supply 9 is used as a backup power supply; effectively avoids damage to the equipment caused by incorrect connection of 380V through the power supply anti-misconnection protection device 10; the present invention has the advantages of simple and reasonable structure, convenient and fast use, realizing real-time state estimation of the substation area, line situation awareness evaluation, and optimizing the substation area maintenance strategy.

[0041] Embodiment 2

[0042] As Figures 1-16 shown, a portable substation area situation awareness analysis device includes a box body 1. A power supply anti-misconnection protection device 10 is installed in the middle of the upper surface of the box body 1. A power supply 2 is slidably installed above the box body 1. An installation frame 3 is arranged in front of the power supply 2. A sensor module 4 is installed on one side of the upper surface of the installation frame 3. A universal adjustable lighting lamp 5 is arranged on the top surface of the sensor module 4; a human-computer interaction mechanism 6 is installed on one side edge of the installation frame 3. A bidirectional communication module 7 is arranged behind the human-computer interaction mechanism 6. The bidirectional communication module 7 is slidably connected to the installation frame 3 and can be retracted into the installation frame 3; a data acquisition and processing module 8 is arranged on one side in front of the installation frame 3. A standby UPS power supply 9 is arranged on the right side of the data acquisition and processing module 8.

[0043] Four bottom pads 11 are provided at the four corners of the bottom of the box body 1. An expansion interface board 12 is provided on the front of the box body 1. The expansion interface board 12 includes a temperature and humidity interface, an infrared interface, an environmental monitoring interface, and other spare interfaces. Slide rails 14 are provided on both sides of the upper surface of the box body 1. A main switch 13 is provided at one corner of the upper surface of the box body 1. A situation awareness analysis module and a fault diagnosis and early warning module are provided inside the box body 1.

[0044] The situation awareness analysis module includes a topology automatic recognition module, a line loss analysis and calculation module, a real-time state estimation module for the power distribution area, a power quality monitoring module, and a maintenance decision optimization module for the power distribution area. The maintenance decision optimization module for the power distribution area includes a line awareness module, a situation assessment module for the power distribution area lines, and a maintenance optimization strategy module for the power distribution area. The line awareness module and the situation assessment module for the power distribution area lines constitute a situation quantization assessment module.

[0045] In this embodiment, the topology automatic recognition module automatically draws a topology diagram of the power supply relationship in the power distribution area through a built-in chip or edge computing technology, and updates the hierarchical relationship in real time (such as transformer - branch - meter box - household meter). The line loss analysis and calculation module uses a high-precision metering chip (0.5s-class active accuracy) to support four-level line loss analysis of the power distribution area, branch, and meter box, and quickly locates abnormal points. The power quality monitoring module supports voltage and current harmonic analysis (2 - 51 times), displays waveforms and harmonic content rate errors (≤0.2%), and provides a data basis 110 for power grid optimization. It also displays the line voltage drop in real time to assist in load balance allocation.

[0046] The real-time state estimation module for the power distribution area uses the graph convolutional network algorithm built in the algorithm library to achieve real-time state estimation of the distribution network based on the spatio-temporal feature graph convolutional network, including the following steps:

[0047] Step 1: Construct a spatio-temporal feature graph convolutional network model: Compose the measured data of the distribution network into a three-dimensional tensor.

[0048] In this embodiment, ① Graph convolutional neural network: For an undirected graph containing n nodes and each node containing v features, at the k-th moment, the features of the entire graph can be represented as a matrix X k ∈R n×v , and the normalized Laplacian matrix L of the graph ∈R n×n is as shown in the following formula: L = I - D -1 / 2 AD -1 / 2 = UΛU T (1), where I is the identity matrix; A is a symmetric adjacency matrix with diagonal elements of 0, and its element a ij = 1 indicates that node i and node j are adjacent, otherwise a ij = 0; D is a diagonal matrix, and its elements satisfy d ij = Π j aij ; U is composed of the eigenvectors of L; Λ is the diagonal matrix of the eigenvalues of L; Since L has the property of real symmetric and positive semi - definite, its eigenvectors have orthogonality, that is, U -1 = U T .

[0049] For the input graph feature X and the convolution kernel θ ∈ R n Define the spectral domain graph convolution operation * G between them, and the calculation process is: θ * G X = U((U T θ) ⊙ (U T X)) = Ug π U T X(2), where ⊙ represents the Hadamard product; g π = diag(U T θ) is the filter. Since the calculation cost of g π is relatively high, the first - order Chebyshev polynomial is used for approximate solution, θ * G X ≈ θ(I + D -1 / 2 AD -1 / 2 )X(3). Since the eigenvalues of (I + D -1 / 2 AD -1 / 2 ) are distributed in [0, 2], numerical instability is introduced during inter - layer transfer. Therefore, consider normalizing it again according to Equation and denote it as where is a diagonal matrix, and its elements satisfy The inter - layer transfer process of the GCN layer is where are the weight matrix, input matrix and output matrix of the l - th layer of the network respectively; c l is the output node feature dimension of the l - th layer; σ(·) is the activation function.

[0050] ② Spatiotemporal feature graph convolution network model: The present invention adopts the STFGCN network, which analyzes and mines the internal correlation information of the spatiotemporal graph from three dimensions of time, space and features, and can be flexibly applied to spatiotemporal graph scenarios such as the state estimation of the distribution network; Based on the spectral - based GCN model, the STFGCN introduces time parameters and bias parameters, and improves the spatial parameters into a learnable full - adjacency matrix. Therefore, the network can comprehensively analyze the multi - dimensional explicit and implicit information of the spatiotemporal graph, so as to achieve tasks such as accurate state estimation.

[0051] Define the graph feature matrix X at time k k as a time slice, and stack c time slices in sequence to form a three - dimensional tensor U k = [X k-c+1 , X k-c+2 ,..., Xk ∈R n×v×c As the input of STFGCN, the inter-layer transfer relationship of the network is expressed as: In the formula, H (0) = U k ; is the spatial parameter, and The initial value matrices expanded by the second dimension are all in formula (4) is the time parameter; is the feature parameter; is the bias parameter; is the tensor mixing multiplier; represents that the sub-matrices obtained by decomposing tensors X and Y along the w-th dimension are multiplied corresponding to each other, and the operation process is as Figure 11 shown; as shown in formula (7), for the operations of spatial and time parameters: In the formula, X and Y are tensors participating in the mixing multiplier operation; Z is the result tensor of the operation; a i , b i , c i respectively represent the dimensions of the i-th dimension of tensors X, Y, and Z.

[0052] When w = 2, there is: In the formula, c1 = a1, c2 = max(a2, b2), c3 = b3. For the operation of feature parameters, there is: When w1 = 1, w2 = 3, there is: In the formula, c1 = max(a1, b1), c2 = b2, c3 = max(a3, b3); Aggregate the spatial information on the spatial axis through left matrix multiplication, Aggregate the time information on the time axis through right matrix multiplication, Aggregate the node attribute information on the feature axis through left matrix multiplication, H (l-1) After feature aggregation and dimension transformation, then superimpose the bias tensor B (l) , and finally obtain the output tensor H through the activation function (l) . The operations of spatial and time parameters satisfy the multiplication associative law, and the ReLU function is selected as the activation function. Specifically, When extracting spatial features, the time dimension is fixed and slides on the node attribute dimension; When extracting time features, the spatial dimension is fixed and slides on the node attribute dimension; When extracting node attribute features, slide on the time and spatial dimensions respectively to achieve multi-scale feature fusion. Compared with the spectral-based GCN, each spatial parameter matrix of STFGCN is a fully adjacent matrix not restricted by the physical topology. Any node in the space can achieve data interaction through only one layer of propagation, and the feature aggregation efficiency is greatly improved.

[0053] Step 2: Construct a real-time state estimation model for the distribution network based on the spatio-temporal feature graph convolutional network: Use the spatio-temporal feature graph convolutional network to extract the feature information of the measurement data on the spatial topology, time series, and node attributes respectively, and obtain the real-time state estimation result through feature fusion;

[0054] Step 3: Generate virtual measurements according to the state estimation results to eliminate the influence of bad data.

[0055] ③ The distribution network state estimation model based on STFGCN includes 2 STFGCN layers, as Figure 12 shown. The network parameters are set as shown in Table 1, where n is the number of nodes in the distribution network, the input time slice length c = 10, that is, the current state is estimated based on the previous 9 measurement time slices and the current measurement time slice; in order to ensure that the spatial information of the tensor does not shift and break during network transmission, the output spatial dimension of the spatial parameter tensor strictly remains the same as the input, that is, n l = n; considering that the state estimation period is short and the state mutation is not obvious, it can be approximately considered that the aggregation relationship of features in space and time is constant in a short time. Therefore, by fixing the spatial dimension and time dimension and simplifying the feature fusion operation of STFGCN, the number of model parameters can be reduced to a certain extent to suppress the overfitting phenomenon; the calculation process of the model is as Figure 13 shown, where k is the current moment and K is the sampling frequency ratio of PMU to SCADA; the historical measurement time slice and the current measurement time slice form a tensor and are input into STFGCN to extract the spatial and time features between the attributes of each node respectively, and then the attributes are fused to obtain the state estimation result. Finally, virtual measurements are generated to correct the defective measurements, completing the measurement fusion link and using it for the state estimation of the next moment.

[0056] Table 1 Parameter settings of the distribution network state estimation model based on STFGCN

[0057]

[0058]

[0059] State estimation: The input node attributes include the node voltage amplitude V i and phase angle θ V,i , injection current amplitude I i and phase angle θ I,i , injection active power P i , injection reactive power Q i . The measured phasor of node i is denoted as The input graph feature matrix is denoted as X k = [X k,1 , X k,2 ,..., Xk,n (12), where each sampling moment of the PMU node has all the measurement values, the SCADA node only has the voltage amplitude and injection power measurements at the SCADA sampling moment, and the measurement values at other moments are all 0; STFGCN will automatically extract the features on the scales of time, space, and node attributes in the measurement time slice sequence, fuse the historical state and measurement information of the system and the current incomplete partial state and measurement data, and output the estimated values of the voltage amplitude and phase angle at the current moment. The output graph feature matrix is For node i, the state estimation value is denoted as

[0060] Virtual measurement: The present invention considers using the STFGCN state estimation result to perform virtual measurement on the current moment time slice to achieve the purpose of repair and denoising, and obtain the virtual measurement of the time slice where the virtual measurement phasor of node i is denoted as The calculation formulas for each item are In the formula, g ij , b ij are respectively the real part and the imaginary part of the element ij of the node admittance matrix; the hat above the variable represents the estimated value of the variable, and the tilde represents the virtual measurement value.

[0061] Initially, the incomplete SCADA measurement values in the measurement time slice X i (i = 1, 2,..., k - 1) are all set to be the same as the last SCADA measurement; at the kth moment, the state estimation value is obtained through the network and the virtual measurement of the time slice is generated At the (k + 1)th moment, use the virtual measurement time slice of the previous moment to replace X k to obtain the input tensor U k+1 , Substitute it into STFGCN to obtain the state estimation result at the (k + 1)th moment; after repeating the above process c - 1 times, the first c - 1 time slices of the input tensor are all the corrected virtual measurements, and the network enters a stable working state. At this time, the input tensor can be denoted as Since the corrected virtual measurement time slice has no missing measurements and has less noise, this model can achieve real-time state estimation at the moment when only PMU measurements are available.

[0062] The situation quantification and evaluation module uses the line situation perception algorithm built in the algorithm library to fully consider the data characteristics and construction status of the advanced metering infrastructure in the distribution area, and uses the impedance calculation model of the main line and branch line of the distribution area users as the line situation perception; it includes the following steps:

[0063] S1: The line perception module calculates the user loop impedance through the user loop impedance calculation module: constructs a user loop impedance equation based on the simplified network topology of the substation area, and uses a constrained linear least squares model for solution;

[0064] In this embodiment, ① User loop impedance calculation: Loop impedance definition: The low-voltage substation area follows the meshed-radiation structure of the power distribution system. Taking single-phase power supply as an example, as Figure 14 shown, the loop impedance of the user is defined as the equivalent impedance of the complete loop starting from the power source, flowing through the user and flowing back to the neutral line of the power source; on this basis, the loop impedance of the user is further differentiated to define the trunk impedance and the branch impedance. The trunk impedance of the user is defined as the equivalent impedance of the line shared with other users; the branch impedance is defined as the impedance of the line exclusively used by the user; as Figure 15 shown, taking User 2 as an example, the power supply loop of User 2 is the part where the branch impedance flows through. The loop impedance of User 2 can be explained as Z Loop(u2) =Z L1 +Z L2 +Z u2 . Obviously, for User 2, the branches where Z L1 and Z L2 are located are shared branches, and the branch where Z u2 is located is the exclusively used branch; so the trunk impedance of User 2 can be expressed as Z GX(u2) =Z L1 +Z L2 , and the branch impedance is Z ZX(u2) =Z u2 , so there is Z Loop(u2) =Z GX(u2) +Z ZX(u2) ;

[0065] Loop impedance model: According to the definitions of the user trunk and branch, write the KVL equation for User 2 in Figure 14 , there is: U T -U2 = Z L1 ×I Σ1 +Z L2 ×I Σ2 +Z u2 ×I u2 =Z GX(u2) ×I GX(u2) +Z ZX(u2) ×I u2 (19), where U T is the voltage of the distribution transformer; I GX(u2) is the equivalent current of the trunk line of User 2, and its magnitude should satisfy I GX(u2) ∈[I Σ2 ,I Σ1, considering the actual situation, the branch impedance of users is mainly reflected in the equivalent impedance of the connection equipment from the power supply in the meter box to the user meter in the actual low-voltage distribution network. The difference of this part for each user is relatively small. From this, it is further deduced to the equivalent current of the main line of any user, and the calculation expression is In the formula, I GX(u2) (k) is the equivalent current of the main line at the k-th moment of a certain day of user 2; g is the user whose daily average voltage in the same phase as user 2 exceeds that of user 2; G is the total number of users in the same phase whose daily average voltage is greater than that of user 2; mean(U g ) is the daily average voltage of user g in the same phase as user 2 under the distribution transformer; mean(U2) is the daily average voltage of user 2; m is the m-th user in the same phase whose daily average voltage does not exceed that of user 2; S is the total number of users in the same phase whose daily average voltage does not exceed that of user 2; I m (k) is the current value at the k-th moment of the m-th user in the same phase whose daily average voltage does not exceed that of user 2.

[0066] Loop impedance calculation: Based on the measurement data of users and distribution transformers at multiple moments every day, according to the expression of formula (19), the following expression can be obtained: In the formula, t1, t2,..., t k are the moments of daily measurement data; obviously, for formula (21), regarding the distribution transformer voltage at each moment as the observed quantity and the main line and branch currents of users as input variables, using the binary regression analysis model to calculate the main line impedance and branch impedance of users. Considering the specific physical meaning of the regression parameters representing the line impedance, the following constraints are provided: Under this constraint condition, formula (21) is fitted to calculate the main line and branch impedances of users.

[0067] S2: Based on the current low-voltage distribution network data and using the user loop impedance calculation as a means, a line state perception and evaluation model for the substation area is proposed;

[0068] S3: Based on the change characteristics of the user loop impedance, the types of line faults in the substation area are classified, and a judgment model for random faults and cumulative fault states of the substation area lines is established to immediately diagnose the abnormal state of the substation area power supply lines.

[0069] In this embodiment, the line state perception and evaluation model for the substation area: By observing the horizontal comparison of the main line and branch impedances of all users in the substation area through the impedance calculation model, the line state of the substation area is evaluated, corresponding to the actual service scenario to support the determination of random line faults, such as loose or broken line connections, etc. Here, the following random fault determination coefficient is designed: stce = isoutlier(Z Loop(k) , `Grubbs′)(23). In the formula, k is the user number; Z Loop(k) = Z GX(k) + Z ZX(k), that is, the total loop impedance of user k is expressed as the sum of the main line impedance and the branch line impedance; Grubbs means using the Grubbs method to determine the calculation of outlier results; isoutlier means performing outlier determination, and the specific criterion follows that when there is at least one user in the substation area whose calculation results are in the outlier state continuously for ω times, it indicates that there is a random anomaly in the substation area line.

[0070] By statistically analyzing the changes in the impedance calculation results of all users in the substation area, a long-term monitoring and analysis model for the substation area line state is established, corresponding to the actual business scenario to support the long-term cumulative deterioration analysis of the line, such as the cracking of the metal conductor surface layer of the line, the thinning of the metal wire due to stress stretching, and the corrosion of the connection points. The designed impedance cumulative fault determination coefficient is: In the formula, d represents the date in season; diff(Z GX(k) (d)) represents the change in the main line impedance of user k at d; N represents the total number of users in the substation area; ltce is the long-term change value of the impedance of the substation area users; when it is found in the calculation results that ltce exceeds a certain set value δ ZK then it is determined that the substation area line state is abnormal and maintenance needs to be arranged in the near future; δ ZK is determined according to the faults caused by the historical cumulative deterioration of the substation area line and its impedance changes; combining the above two points, the maintenance trigger function TF can be obtained as the function of the minimum number of consecutive outlier times ω of the single-user impedance and the continuous impedance deviation threshold δ ZK , that is, TF = f(ω, δ ZK )(25).

[0071] The substation area optimized maintenance strategy module uses the non-linear programming algorithm built in the algorithm library to construct an optimized maintenance strategy for the substation area with sufficient maintenance costs; including the following steps:

[0072] P1: Construct an operation effectiveness evaluation model considering maintenance costs: Combining the maintenance costs and revenue methods of the substation area, a calculation model for the operation effectiveness of the substation area is constructed with the goal of maximizing the operation effectiveness;

[0073] P2: Establish an optimized maintenance strategy for the substation area: Combining the power supply reliability constraint conditions, establish an optimized maintenance decision model for the substation area based on the non-linear programming method, and solve the parameters of the optimized line state determination model through the constrained non-linear programming method to guide the maintenance work of the substation area line state, optimize the maintenance process, maximize the effectiveness of maintenance, and improve the operation effectiveness.

[0074] In this embodiment, the optimized maintenance strategy for the substation area: The traditional maintenance cost of the substation area consists of two parts: maintenance cost and load loss, that is, S MD = S MC + S OC , where S MC is the maintenance cost; S OCLosses caused by power outages during the period from fault power outages to the restoration of power supply after maintenance is completed; S MC , S OC Both are related to the maintenance type MT ∈ {major overhaul, emergency repair, minor repair, inspection}. Calculating the annual operation and maintenance cost on an annual basis can be expressed as In the formula, FC is the set of annual maintenance events; i is the serial number of the maintenance event in the year; q i is the maintenance type of the i-th maintenance event q i ∈ MT.

[0075] Cost of condition-based maintenance losses: There will be a certain time difference between the time when condition-based maintenance is triggered and the specific time when maintenance is actually carried out. Define the condition-based maintenance response degree to measure this time difference, that is In the formula, τ is the response degree of condition-based maintenance, and its value ranges from 0% to 100%; t act is the time when maintenance is carried out; t trigger_on is the time when the trigger function is triggered; t trigger_off is the time when the trigger function trigger disappears; as Figure 16 shown, the condition-based maintenance losses of the distribution transformer area calculated based on the loop impedance have three stage characteristics:

[0076] In the first stage, the condition-based maintenance trigger function is triggered. After a period of time, the line deterioration is relatively slow, and the necessary investment required for maintenance also increases relatively slowly.

[0077] In the second stage, after the condition-based maintenance is triggered, the maintenance is not carried out in a timely manner, and the state of the line deteriorates sharply, resulting in a sharp increase in the necessary investment for maintenance.

[0078] In the third stage, the line deterioration has stabilized. At this time, the necessary investment in condition-based maintenance has stabilized, and only the losses caused by power outages increase with time.

[0079] Therefore, under the condition-based maintenance method of the distribution transformer area, the annual operation and maintenance losses can be expressed as In the formula, γ is the type of line abnormality that triggers the condition-based maintenance event, γ ∈ {stce, ltce}; g(γ, τ, t trigger_on , t trigger_off ) represents the single maintenance loss calculation function with the abnormality type, maintenance response degree, trigger function trigger time, and trigger function trigger disappearance time as variables.

[0080] Analysis of the operation effectiveness of the distribution transformer area: The operation effectiveness of the distribution transformer area is reflected in the economic benefits generated by the stable power supply of the distribution transformer area minus the operation and maintenance costs of the distribution transformer area, that is In the formula, S ME is the operation effectiveness; S PSR is the benefit generated by power supply, and its magnitude is affected by the uninterrupted power supply duration and load of the distribution transformer area; S CBMThe maintenance loss under the condition-based maintenance mode; the power supply duration of the power distribution area is affected by the operation and maintenance level of the power distribution area. On the premise that the necessary investment in dealing with various faults is relatively fixed, the operation effect of the power distribution area line under the condition-based maintenance strategy based on loop impedance calculation can be summarized as an optimization problem, that is, maxS ME = S PSR - S CBM (31). To optimize the solution of the operation effect, constraint conditions need to be added, and the calculation expression becomes In the formula, H is the total number of users; t power_off (k) is the annual power outage duration of user k; T Limit is the limit value of the average annual power outage duration per household stipulated by the State Grid. According to the above analysis, by solving the non-linear programming problem under the calculation constraints, the appropriate parameters ω and δ ZK are determined, and directly adjust the trigger frequency of TF = f(ω, δ ZK ), further affecting the maintenance execution frequency, and ultimately affecting the power supply duration and maintenance loss; therefore, optimizing and selecting reasonable ω and δ ZK can obtain the maximum operation effect.

[0081] In summary, the present invention aims at the problem of the condition-based maintenance strategy of low-voltage power distribution area lines, based on the calculation of the impedance of low-voltage power distribution area lines, deeply combines the current construction status of the distribution network business system, constructs a calculation model for the impedance of user loops in low-voltage power distribution areas, analyzes the health status of the lines in real time, and innovatively proposes a random fault and cumulative fault discrimination model based on the analysis of the impedance state of user loops, effectively identifying the random anomalies and cumulative deterioration problems of power distribution area lines.

[0082] The present invention relates to a portable power distribution area situation awareness analysis device. During use, the real-time state estimation module of the power distribution area of the present invention is based on real-time state estimation of the distribution network using a spatio-temporal feature graph convolutional network. That is, in a scenario considering the spatio-temporal characteristics of node loads and incomplete measurements, a real-time state estimation algorithm for the distribution network based on a spatio-temporal feature graph convolutional network (STFGCN) is proposed. 1) Model the state estimation problem of the distribution network from a graph perspective. The STFGCN network can adaptively aggregate information in three dimensions: the spatial topology, time series, and node attributes of the spatio-temporal graph of the distribution network, thereby achieving accurate state estimation. 2) It can generate high-quality virtual measurements in extreme scenarios with incomplete measurements to eliminate the influence of bad data. The robustness and computational speed of the algorithm are significantly better than traditional methods (it can automatically complete the multi-time scale matching of measurement data and the multi-feature fusion of spatio-temporal graphs, avoiding the influence of bad data, and can achieve state estimation with only a small number of PMU measurements). 3) When processing time series data, compared with common neural networks such as multi-layer perceptron (MLP), recurrent neural network (RNN), and long short-term memory network (LSTM), the parameter scale of STFGCN can be significantly compressed, and it has stronger interpretability and generalization ability. The present invention has the advantages of simple and reasonable structure, convenient and fast use, realizing real-time state estimation of the power distribution area, line situation awareness assessment, and optimizing the maintenance strategy of the power distribution area.

Claims

1. A portable substation area situation awareness and analysis device, which comprises a box body, and is characterized in that: A power supply anti-misconnection protection device is installed in the middle of the upper surface of the box body. A power supply is slidably installed above the box body. An installation frame is arranged in front of the power supply. A sensor module is installed on one side of the upper surface of the installation frame. A universal adjustable lighting lamp is arranged on the top surface of the sensor module. A man-machine interaction mechanism is installed on one side edge of the installation frame. A two-way communication module is arranged behind the man-machine interaction mechanism. The two-way communication module is slidably connected to the installation frame and can be telescoped into the installation frame. On one side of the front of the installation frame, a data acquisition and processing module is arranged. A standby UPS power supply is arranged on the right side of the data acquisition and processing module.

2. The portable substation area situation awareness and analysis device according to claim 1, characterized in that: Bottom pads are arranged at the four corners of the bottom of the box body. An expansion interface board is arranged on the front of the box body. The expansion interface board includes a temperature and humidity interface, an infrared interface, an environmental monitoring interface, and other standby interfaces. Slide rails are arranged on both side edges of the upper surface of the box body. A main switch is arranged at one corner of the upper surface of the box body. A situation awareness analysis module and a fault diagnosis and early warning module are arranged inside the box body.

3. The portable substation situation awareness analysis device according to claim 2, characterized in that: The power supply includes a power module. A bottom plate is arranged at the bottom of the power module. T-shaped sliding plates that are slidably connected to the corresponding slide rails are arranged on both sides of the lower surface of the bottom plate.

4. The portable substation situation awareness analysis device according to claim 3, characterized in that: The sensor module is used to collect data of the distribution area, including a current transformer, a voltage transformer, a power sensor, an image sensor, and a smoke sensor.

5. The portable substation area situation awareness and analysis device according to claim 4, wherein: The man-machine interaction mechanism includes a C-shaped frame arranged on the side edge of the installation frame. A rotating seat is arranged at the free end above the C-shaped frame. A man-machine interaction large screen for providing an operation interface and data display is arranged inside the rotating seat. The rotating seat is movably connected to one side edge of the man-machine interaction large screen through a universal ball seat device.

6. The portable substation situation awareness analysis device according to claim 5, wherein: The two-way communication module is used to realize data transmission between the device and an external system, including a two-way communicator. An outer frame board is installed outside the two-way communicator. A handle is arranged at the top of the outer frame board. Hinged boards are hingedly installed on both side surfaces at the lower ends of the outer frame board. The hinged boards are slidably connected to a sliding channel located inside the installation frame. A heat dissipation port is arranged on the outer side surface of the data acquisition and processing module. A heat dissipation fan is installed inside the heat dissipation port. The data acquisition and processing module is used to process and analyze the data of the sensor module, including a real-time data acquisition module, a data normalization processing module, and an algorithm library.

7. The portable substation situation awareness and analysis device according to claim 2, characterized in that: The situation awareness analysis module includes a topology automatic recognition module, a line loss analysis and calculation module, a distribution area real-time state estimation module, a power quality monitoring module, and a distribution area maintenance decision optimization module. The distribution area maintenance decision optimization module includes a line awareness module, a distribution area line situation assessment module, and a distribution area optimized maintenance strategy module. The line awareness module and the distribution area line situation assessment module constitute a situation quantification assessment module.

8. The portable substation situation awareness analysis device according to claim 7, wherein: The distribution area real-time state estimation module uses the graph convolutional network algorithm built in the algorithm library to realize the real-time state estimation of the distribution network based on the spatio-temporal feature graph convolutional network, including the following steps: Step 1: Construct a spatio-temporal feature graph convolutional network model: form a three-dimensional tensor from the distribution network measurement data; Step 2: Construct a real-time state estimation model for the distribution network based on the spatio-temporal feature graph convolutional network: Use the spatio-temporal feature graph convolutional network to extract the feature information of the measurement data in terms of spatial topology, time series, and node attributes respectively, and obtain the real-time state estimation result through feature fusion; Step 3: Generate virtual measurements according to the state estimation result to eliminate the influence of bad data.

9. The portable substation situation awareness analysis device according to claim 7, wherein: The situation quantification and evaluation module uses the line situation perception algorithm built in the algorithm library to realize fully considering the data characteristics and construction status of the advanced metering infrastructure in the substation area, and uses the impedance calculation model of the main line and branch line of the users in the substation area as the line situation perception; it includes the following steps: S1: The line perception module calculates the impedance of the user loop through the user loop impedance calculation module: Construct a user loop impedance equation based on the simplified network topology of the substation area, and use a constrained linear least squares model for solution; S2: Based on the current low-voltage distribution network data and taking the calculation of the user loop impedance as a means, propose a substation area line state perception and evaluation model; S3: Based on the change characteristics of the user loop impedance, classify the fault types of the substation area lines, establish a random fault and cumulative fault state judgment model for the substation area lines, and instantly diagnose the abnormal state of the substation area power supply lines.

10. A portable substation situation awareness and analysis device according to claim 7, characterized in that: The substation area optimized maintenance strategy module uses the nonlinear programming algorithm built in the algorithm library to construct an optimized maintenance strategy for the substation area that fully considers the maintenance cost; it includes the following steps: P1: Construct an operation effectiveness evaluation model considering the maintenance cost: Integrate the maintenance cost and revenue methods of the substation area, and construct a substation area operation effectiveness calculation model with the maximization of operation effectiveness as the goal; P2: Establish an optimized maintenance strategy for the substation area: Combine the power supply reliability constraint conditions, establish an optimized substation area maintenance decision-making model based on the nonlinear programming method, and solve the parameters of the optimized line state judgment model through the constrained nonlinear programming method to guide the state maintenance work of the substation area lines and optimize the maintenance process.

Images