A smart box-type transformer control system

Abstract

The present invention discloses a smart box-type transformer control system, which includes a smart box-type transformer distribution structure, a synchronous phasor measurement unit, a dynamic reactive power support unit, and a heavy overload control unit. Compared with the traditional smart box-type transformer control system, the present invention can solve the problems of weak dynamic reactive power support capability on the load side and prominent heavy overload problems during peak load periods in the traditional smart box-type transformer control system, large amount of calculation and low calculation accuracy when facing a new power system with extensive access to power electronic equipment, and has guiding significance for the design of smart box-type transformer control systems.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution of electric power systems, and in particular to an intelligent box-type transformer control system. Background Art

[0002] Smart box-type transformer is a prefabricated distribution device that integrates distribution transformers and their high and low voltage side switches as well as intelligent monitoring equipment and control systems. It has many advantages such as small footprint, easy installation, compact equipment, and high intelligence. As the terminal component of the power system, smart box-type transformer plays an important role in the transmission of electricity. As the core component of smart box-type transformer, the design and implementation of its control system is particularly important.

[0003] The existing control systems of smart box transformers are mostly based on the equipment itself, focusing more on the data collection and monitoring of the equipment, lacking automatic control of the equipment, and the actual operation process of the smart box transformer often depends on the remote equipment status operation of the power grid control personnel. At the same time, as an important link in the power production chain, the existing control ideas of smart box transformers often cannot take into account the actual operation of the power grid. There are shortcomings such as weak dynamic reactive power support capacity on the load side and prominent heavy overload problems during peak load periods. With the access of low-voltage distributed renewable energy and power electronic equipment, the traditional idea of ​​providing reactive power support and heavy overload prediction based on the power system flow calculation method faces the shortcomings of large calculation amount and low accuracy of high-dimensional nonlinear numerical solutions, which will have a significant adverse impact on the development of new energy, the improvement of power quality on the user side, and the safe and stable operation of box transformers. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to provide a smart box transformer control system, so that the smart box transformer can better consider the voltage quality on the load side, equipment overload conditions and other operating information during the actual operation process, improve the support capacity and operation reliability of the smart box transformer for the power grid, and be closer to the actual operation of the power system.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a smart box-type transformer control system, which includes a smart box-type transformer distribution structure, a synchronous phasor measurement unit, a dynamic reactive power support unit, and a heavy overload control unit;

[0007] The smart box-type transformer distribution structure is a centralized 35kV distribution system, which consists of a 35kV bus, a 35kV main transformer, a 10kV main line, a 10kV branch line and a reactive compensation device. Each of the distribution system structures is equipped with a signal sensor for on-site measurement of actual operating data.

[0008] The synchronized phasor measurement unit is used to collect the voltage, current phasor and grid frequency of each distribution system component structure, and to send actual operation data to the dynamic reactive support unit and the heavy overload control unit in a data-driven manner through communication;

[0009] The dynamic reactive power support unit is used to receive the operation data collected by the synchronous phasor measurement unit, and input the data set into the optimal network structure after multi-layer unsupervised reinforcement learning, and dynamically optimize the switching amount of the main transformer tap, capacitor group, and reactor group through the output dynamic reactive power optimization variables;

[0010] The heavy overload control unit is used to receive the operating data collected by the synchronous phasor measurement unit, and input the data set into the optimal network structure after multi-layer unsupervised reinforcement learning. By predicting the heavy overload situation at a future moment, the line is stopped according to the power limitation sequence table, and a closing command is issued to the opposite side of the open-loop line.

[0011] In a second aspect, the present invention provides a dynamic reactive power optimization algorithm for a smart box transformer, comprising the following steps:

[0012] A1. Collect the actual operation data of each component of the smart box-type transformer and distribution system, and send the collected actual operation data to the dynamic reactive power support unit;

[0013] A2. The dynamic reactive power support unit conducts unsupervised reinforcement learning training on the model through a large sample training set and an actual test set to build a dynamic reactive power optimization model;

[0014] A3. The dynamic reactive power support unit normalizes the actual operation data as the input of the dynamic reactive power optimization model, and obtains the dynamic reactive power optimization result through input-output mapping;

[0015] A4. Send the dynamic reactive power optimization results as input instructions to the main transformer tap position controller, capacitor bank, and reactor group switching switches to achieve real-time dynamic reactive power optimization;

[0016] Furthermore, the specific process of A2 is as follows: a large sample training set and a test set with a ratio of 9:1 are generated through Monte Carlo sampling, and the training set and the test set are input into an extreme learning machine for unsupervised reinforcement learning training, and finally a dynamic reactive power optimization model is generated.

[0017] In a third aspect, the present invention provides a heavy overload control algorithm for a smart box-type transformer, comprising the following steps:

[0018] B1. Collect the actual operation data of each component of the smart box-type transformer and distribution system, and send the collected actual operation data to the heavy overload control unit;

[0019] B2. The heavy overload control unit conducts unsupervised reinforcement learning training on the model through a large sample training set and a test set to build a power flow prediction model;

[0020] B3. The heavy overload control unit normalizes the actual operation data as the input of the power flow prediction model and obtains the power flow prediction result through input-output mapping;

[0021] B4. The heavy overload control unit will set the power flow prediction value η s With the main transformer heavy overload capacity h s , predict heavy overload time t s Make a comprehensive judgment, if η s >1.7h s , and t s If the load exceeds 30 minutes, the current is judged to be out of limit, and the heavy overload control unit sends an opening command to the line and a closing command to the opposite side of the open-loop line at the same time.

[0022] Furthermore, the specific process of S2 is as follows: a large sample training set and a test set with a ratio of 9:1 are formed by a Monte Carlo sampling data set and a historical data set, and the training set and the test set are input into an extreme learning machine for unsupervised reinforcement learning training, and finally a power flow prediction model is generated.

[0023] Furthermore, the specific steps of the heavy overload control measures in the above S4 are: after receiving the power flow exceeding the limit information, the heavy overload control unit reads the power limiting priority table, and sends an opening instruction to the circuit breaker on the line side in sequence according to the power limiting order in the power limiting priority table, and at the same time sends a closing instruction to the opposite side of the open-loop line until the heavy overload control unit no longer generates power flow exceeding the limit information.

[0024] The present invention adopts the above technical solution, which has the following advantages:

[0025] 1. The present invention provides a smart box-type transformer control system with a clear structure. In the actual operation process, it can better consider the voltage quality on the load side, the heavy overload condition of the equipment and other operating information, improve the supporting capacity and operating reliability of the smart box-type transformer for the power grid, and be closer to the actual operation of the power system.

[0026] 2. The dynamic reactive power optimization algorithm of the smart box-type transformer proposed in the present invention is suitable for new power systems with a high proportion of distributed renewable energy and a high proportion of power electronic equipment connected. Compared with the traditional reactive power optimization combination obtained by the Newton-Raphson method flow calculation method, it has the characteristics of high calculation accuracy and fast solution speed.

[0027] 3. The heavy overload control algorithm of the intelligent box-type transformer proposed in the present invention adopts a power flow prediction model and a heavy overload equipment control idea based on the power limit priority table, which is suitable for new power systems with a high proportion of distributed renewable energy and a high proportion of power electronic equipment connected. It has the characteristics of high prediction accuracy, fast solution speed and high power supply reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Throughout the accompanying drawings, the same reference numerals are used to represent the same components. In the accompanying drawings:

[0029] Figure 1 This is the structure diagram of the intelligent box-type transformer control system of the present invention;

[0030] Figure 2 This is a flow chart of the dynamic reactive power optimization algorithm of the smart box-type transformer of the present invention;

[0031] Figure 3 It is a flow chart of the intelligent box variable weight overload control algorithm of the present invention. DETAILED DESCRIPTION

[0032] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0033] It should be understood that the terms used herein are only for the purpose of describing specific example embodiments and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "include", "comprise" and "have" are inclusive and therefore specify the existence of stated features, steps, operations, elements and / or parts, but do not exclude the existence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not interpreted as necessarily requiring them to be performed in the specific order described or illustrated, unless the execution order is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0034] This embodiment provides a smart box-type transformer control system, such as Figure 1As shown. It includes a smart box transformer distribution structure, a PMU (synchronous phasor measurement unit) whose acquisition points are adapted to the smart box transformer distribution structure, a dynamic reactive power support unit, and a heavy overload control unit.

[0035] The smart box-type transformer distribution structure in this embodiment is a centralized 35kV distribution system, which consists of a 35kV bus, a 35kV main transformer, a 10kV main line, a 10kV branch line, capacitor groups 1 to 4 and reactor groups, and feeder lines 1 to 8. After being transmitted to the 35kV bus through a higher voltage level power system, the electric energy enters the centralized smart box-type transformer and is finally delivered to the user.

[0036] The synchronous phase measurement unit mainly measures the voltage, current phasor and grid frequency data at different measuring points inside the smart box transformer, and synchronously sends the real-time measurement data to the dynamic reactive support unit and the heavy overload control unit. It also receives feedback signals from the above two units for data collection.

[0037] The dynamic reactive power support unit plays a role in voltage support and power quality improvement in the intelligent box-type transformer control system. The calculation part in the dynamic reactive power support unit will generate an optimal network 1 after multi-layer unsupervised reinforcement learning training. After receiving the measurement data of the synchronous vector measurement unit, the optimal network 1 will automatically generate a dynamic reactive power optimization instruction, thereby achieving the role of voltage support by controlling the switching amount of capacitor groups 1 to 4 and reactor groups and the tap position of the main transformer and other variables.

[0038] The heavy overload control unit plays a role in power flow control in the intelligent box transformer control system to ensure the safe operation of the main transformer. The calculation part in the heavy overload control unit will generate the optimal network 2 after multi-layer unsupervised reinforcement learning training. When the optimal network 2 inputs data, it will generate the over-limit power flow value and over-limit time of the main transformer. If the over-limit power flow value and over-limit time of the power flow exceed the control range, the circuit breaker opening command will be issued to the feeder after reading the power limit sequence table, so as to control and intervene the power flow of the main transformer off-grid to ensure that the main transformer operates in a safe and stable range.

[0039] like Figure 2 As shown, the dynamic reactive power optimization algorithm provided in this embodiment includes the following contents:

[0040] A1. Construct reactive power optimization model;

[0041] Specifically, Figure 2 As shown, the training set and test set with a sample ratio of 9:1 generated by Monte Carlo sampling are input into the unsupervised reinforcement learning trainer, and the number of neuron nodes in each layer of the generated neural network is determined by the training effect deviation rate as an indicator, and finally the reactive power optimization model is determined.

[0042] In a specific embodiment of the present invention, the sample number of the Monte Carlo sampling algorithm adopted by the present invention is n=10000, the training accuracy (goal) is 0.001, the maximum number of verification failures (max fail) is 20 times, and the matrix of the number of neuron nodes in each layer obtained thereby is [30, 120, 70, 5].

[0043] A2. Dynamic reactive power optimization

[0044] Specifically, Figure 2 As shown, the operating data is input into the reactive power optimization model determined in A1 after normalization, and its input-output function is:

[0045] h(C,L,p)=ψ(i,u)

[0046] Among them, C and L are the switching values ​​of capacitor groups No. 1 to No. 4 and reactor groups respectively, p is the tap position of the main transformer, i and u are the current and voltage phasors of each collection point respectively. The reactive power optimization result is substituted into the smart box transformer distribution structure for flow calculation, and the voltage value of each node is output.

[0047] A3. Dynamic reactive power optimization command output

[0048] Specifically, Figure 2 As shown, the voltage value of each node is compared with the voltage reference value to determine whether it exceeds the limit. In the embodiment of the present invention, according to the power system regulations, the range of the node voltage matrix is ​​selected to be 0.97U s <U<1.0U7 s Among them, U s It is the voltage reference value corresponding to the corresponding voltage level.

[0049] If the node voltage exceeds the voltage reference value and its over-limit ratio, the reactive power optimization command will not be output. The input data will be recalculated for dynamic reactive power optimization. If the node voltage does not exceed the voltage reference value and its over-limit ratio, the reactive power optimization matrix [C, L, p] calculated in A2 will be output to the controller to control the switching amount of capacitor groups 1 to 4 and reactor groups and the tap of the main transformer.

[0050] like Figure 3 As shown, the heavy overload control algorithm provided in this embodiment includes the following contents:

[0051] B1. Constructing a heavy overload flow prediction model

[0052] Specifically, Figure 3As shown in the figure, the samples generated by Monte Carlo sampling and the historical data samples are used to form a training set and a test set with a ratio of 9:1, which are input into the unsupervised reinforcement learning trainer. The number of neuron nodes in each layer of the generated neural network is determined based on the training effect deviation rate, and the heavy overload power flow prediction model is finally determined.

[0053] In a specific embodiment of the present invention, the number of samples of the Monte Carlo sampling algorithm adopted by the present invention is n=7000, the number of samples of historical trend data is n=3000, the training accuracy (goal) is 0.001, the maximum number of verification failures (maxfail) is 20 times, and the matrix of the number of neuron nodes in each layer obtained thereby is [40,130,60,4].

[0054] B2. Heavy overload logic judgment

[0055] Specifically, Figure 3 As shown in Figure 1, the operating data is normalized and input into the heavy overload power flow prediction model determined in B1, and its input-output function is:

[0056] p(η s ,t s )=ξ(i,u)

[0057] The output result is the long-term power flow prediction value η of the main transformer in the future s and prediction time t s . Set the value η s With the main transformer heavy overload capacity h s Make a comprehensive judgment, if η s >1.7h s , then proceed to the next step of over-limit time judgment. s <1.7h s , then the return command calls the running data input power flow prediction model again. In the above steps, if t s If the current exceeds the limit, the control command will be output.

[0058] B3. Execution of power flow over-limit instruction

[0059] Specifically, Figure 3 As shown, after receiving the power flow over-limit control instruction, the heavy overload control unit will call the built-in power limiting priority table, and send an opening instruction to the feeder switches No. 1 to 8 according to the priority of the power limiting priority table. At the same time, if the open-loop control word of the line in the power limiting priority table is 1, a closing instruction will be sent to the switch on the opposite side.

[0060] In a specific embodiment of the present invention, the power restriction sequence table is built into the control program of the host computer of the intelligent box-type transformer control system, and is updated in real time when the line is put into operation or retired, and when the user-side information changes. After the line operation mode is adjusted, the operator will also modify the open-loop control word according to the actual open-loop and closed-loop operation of the line.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A smart box-type transformer control system, characterized by: It includes intelligent box-type transformer distribution structure, synchronous phasor measurement unit, dynamic reactive power support unit and heavy overload control unit; The smart box-type transformer distribution structure is a centralized 35kV distribution system, which consists of a 35kV bus, a 35kV main transformer, a 10kV main line, a 10kV branch line and a reactive compensation device. Each of the distribution system structures is equipped with a signal sensor for on-site measurement of actual operating data. The synchronized phasor measurement unit is used to collect the voltage, current phasor and grid frequency of each distribution system component structure, and transmit the actual operation data to the dynamic reactive power support unit and the heavy overload control unit in a data-driven manner through communication; The dynamic reactive power support unit is used to receive the operation data collected by the synchronous phasor measurement unit, and input the data set into the network structure after multi-layer unsupervised reinforcement learning, and dynamically optimize the switching amount of the main transformer tap, capacitor group, and reactor group through the output dynamic reactive power optimization variables; The heavy overload control unit is used to receive the operation data collected by the synchronous phasor measurement unit, and input the data set into the network structure after multi-layer unsupervised reinforcement learning, and stop the line according to the power limitation sequence table by predicting the heavy overload situation at the future moment, and at the same time send a closing command to the opposite side of the open-loop line; It also includes a dynamic reactive power optimization algorithm for a smart box-type transformer, which specifically includes the following steps: A1. Collect the actual operation data of each component of the smart box-type transformer and distribution system, and send the collected actual operation data to the dynamic reactive power support unit; A2. The dynamic reactive power support unit conducts unsupervised reinforcement learning training on the model through a large sample training set and an actual test set to build a dynamic reactive power optimization model; A3. The dynamic reactive power support unit normalizes the actual operation data as the input of the dynamic reactive power optimization model, and obtains the dynamic reactive power optimization result through input-output mapping; A4. Send the dynamic reactive power optimization results as input instructions to the main transformer tap position controller, capacitor bank, and reactor bank switching switches to achieve real-time dynamic reactive power optimization.

2. The intelligent box-type transformer control system according to claim 1 is characterized in that: The specific process of step A2 is as follows: Monte Carlo sampling is used to generate a large sample training set and a test set with a ratio of 9:1, and the training set and the test set are input into an extreme learning machine for unsupervised reinforcement learning training, and the number of neuron nodes in each layer of the generated neural network is determined by taking the training effect deviation rate as an indicator, and finally a dynamic reactive power optimization model is generated.

3. The intelligent box-type transformer control system according to claim 1 is characterized in that: The invention also includes a heavy overload control algorithm of a smart box-type transformer, which specifically includes the following steps: B1. Collect the actual operation data of each component of the smart box-type transformer and distribution system, and send the collected actual operation data to the heavy overload control unit; B2. The heavy overload control unit conducts unsupervised reinforcement learning training on the model through a large sample training set and a test set to build a power flow prediction model; B3. The heavy overload control unit normalizes the actual operation data as the input of the power flow prediction model and obtains the power flow prediction result through input-output mapping; B4. The heavy overload control unit makes a comprehensive judgment on the predicted value of the main transformer power flow, the main transformer heavy overload capacity, and the predicted heavy overload time. The heavy overload control unit sends an opening command to the line and a closing command to the opposite side of the open-loop line at the same time.

4. The intelligent box-type transformer control system according to claim 3 is characterized in that: The specific process of step B2 is: a large sample training set and a test set with a ratio of 9:1 are formed by a Monte Carlo sampling data set and a historical data set, and the training set and the test set are input into an extreme learning machine for unsupervised reinforcement learning training, and finally a power flow prediction model is generated.

5. The intelligent box-type transformer control system according to claim 3 is characterized by: The specific process of step B4 is: Heavy overload capacity of main transformer , predict heavy overload time Make a comprehensive judgment, if ,and >30min, the power flow is judged to be out of limit. After receiving the power flow out of limit information, the heavy overload control unit reads the power limit priority table, and sends opening instructions to the circuit breaker on the line side in sequence according to the power limit sequence in the power limit priority table, and sends closing instructions to the opposite side of the open-loop line at the same time.

Images