Black start system and method for diesel combined combustion engine

Abstract

To provide a black start system and method for a diesel combined combustion engine.SOLUTION: A method includes: obtaining the closing overvoltage and multiple voltage influence parameters of a no-load line in the historical black-start of a power grid; calculating a first weight set and a second weight set; determining a first error and a second error based on the closing overvoltage, the multiple voltage influence parameters, the first weight set and the second weight set; updating the first weight set and the second weight set according to a predetermined number of iterations to obtain a first weight optimized set and a second weight optimized set; obtaining a closing overvoltage target prediction value set based on a target weight set and multiple real-time voltage influence parameters in the current black start; and supplying power to a power grid to complete the black start of the power grid.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present application relates to the technical field of black start overvoltage prediction, and in particular to a black start system and method for diesel combined combustion engines. [Background technology]

[0002] Large-scale power outages occurring in recent years are often caused by potential hazards in the power grid, with specific triggering factors causing the grid to collapse or collapse. Black start refers to the process of starting up self-starting units within the system without relying on other networks after a power grid or entire system shuts down due to a fault, thereby driving the startup of units that do not have self-starting capabilities, gradually expanding the recovery range of the power system and ultimately achieving the recovery of the entire power system. However, existing black start systems have poor voltage stability during startup and take a long time to start up, preventing the local power grid from operating safely and stably.

[0003] Conventional techniques have been proposed to rapidly predict no-load closing overvoltages using artificial intelligence. However, there are many factors that affect black start overvoltages (i.e., closing overvoltages on no-load lines), and conventional techniques do not take into account the effects of factors such as line parameters and the initial phase angle of the circuit breaker closing circuit on overvoltages, resulting in inaccurate prediction results. Furthermore, they do not take into account the differences in the degree to which different factors affect overvoltages, which also affects the prediction results. Summary of the Invention

[0004] The present application aims to solve at least part of the technical problems in the related art.

[0005] Therefore, the first objective of the present application is to provide a black start method for a diesel combined combustion engine, which can better predict the closing overvoltage of different starting paths and select the optimal path.

[0006] A second object of the present application is to provide a black start system for a diesel combined combustion engine.

[0007] A third object of the present application is to provide an electronic device.

[0008] A fourth object of the present application is to provide a computer-readable recording medium.

[0009] In order to achieve the above object, an embodiment of the first aspect of the present application proposes a black start method for a diesel combined combustion engine, in which a diesel black start unit is arranged, and the diesel black start unit is used to start the gas generator through a starting device of the gas generator when the gas generator stops due to a loss of power in the power grid, and the black start method includes: obtaining a closing overvoltage and a plurality of voltage influence parameters of an unloaded line when the power grid has previously experienced a black start; calculating a first set of weights and a second set of weights for the closed-circuit overvoltage and the plurality of voltage-influence parameters using Pearson correlation coefficient and cosine similarity; determining a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage-influencing parameters, the first set of weights, and the second set of weights using a recurrent neural network model; updating the first weight set and the second weight set according to the first error and the second error according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set; using a recurrent neural network model to obtain a set of closed-circuit overvoltage target prediction values ​​based on the set of target weights and a plurality of real-time voltage influence parameters in the current black start; and supplying power to the power grid to complete the black start of the power grid by selecting the line on which the minimum value in the set of closed-circuit overvoltage target prediction values ​​is located as the optimal line.

[0010] In the method of the first aspect of the present application, calculating a first weight set and a second weight set for the closed-circuit overvoltage and the plurality of voltage-influence parameters using the above-mentioned Pearson correlation coefficient and cosine similarity includes: calculating a first correlation coefficient set for the closed-circuit overvoltage and the plurality of voltage-influence parameters using the Pearson correlation coefficient; obtaining a first weight set based on the sum of the absolute value of each first correlation coefficient in the first correlation coefficient set and the absolute values ​​of all the first correlation coefficients; calculating a second correlation coefficient set for the closed-circuit overvoltage and the plurality of voltage-influence parameters using the cosine similarity; and obtaining a second weight set based on the sum of the absolute value of each second correlation coefficient in the second correlation coefficient set and the absolute values ​​of all the second correlation coefficients.

[0011] In the method of the first aspect of the present application, determining a first error and a second error based on the closed circuit overvoltage, the plurality of voltage-influencing parameters, the first weight set, and the second weight set using the above-mentioned recurrent neural network model includes: obtaining first model input data based on the plurality of voltage-influencing parameters and the first weight set, inputting the first model input data into a recurrent neural network model to obtain a first set of overvoltage prediction values, and obtaining a first error based on the closed circuit overvoltage and the first set of overvoltage prediction values; obtaining second model input data based on the plurality of voltage-influencing parameters and the second weight set, inputting the second model input data into a recurrent neural network model to obtain a second set of overvoltage prediction values, and obtaining a second error based on the closed circuit overvoltage and the second set of overvoltage prediction values.

[0012] In the method of the first aspect of the present application, updating the first weight set and the second weight set according to a predetermined number of iterations based on the above-mentioned first error and second error to obtain the first weight optimization set and the second weight optimization set, and further obtaining the target weight set includes determining whether the current number of iterations is equal to the predetermined number of iterations, and if not, comparing the first error and the second error, and updating the first weight set and the second weight set according to different requirements based on the comparison result until the current number of iterations is equal to the predetermined number of iterations, and obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set under each number of iterations.

[0013] In the method of the first aspect of the present application, updating the first weight set and the second weight set based on the above-mentioned first error and second error according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set further includes using a recurrent neural network model to determine a first error target value and a second error target value based on the closed circuit overvoltage, the plurality of voltage influence parameters, and the first weight optimization set and the second weight optimization set, and selecting the weight optimization set corresponding to the smallest value of the first error target value and the second error target value as the target weight set.

[0014] In the method of the first aspect of the present application, updating the first weight set and the second weight set according to different requirements based on the comparison result until the above-mentioned current number of iterations is equal to the predetermined number of iterations includes: if the first error is greater than the second error, updating the first weight set and the second weight set according to a first step length and a second step length, respectively; and if the first error is less than or equal to the second error, updating the first weight set and the second weight set according to a first proportion and a second proportion, respectively.

[0015] In the method of the first aspect of the present application, obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set under each of the above-mentioned iteration numbers includes obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set in the final iteration and the first weight set and the second weight set under the iteration number corresponding to the case where the sum of the first error and the second error is smallest.

[0016] In order to achieve the above object, an embodiment of a second aspect of the present application proposes a black start system for a diesel combined combustion engine, wherein a diesel black start unit is arranged, and the diesel black start unit is used to start a gas generator through a starting facility of the gas generator when the gas generator stops due to a loss of power in the power grid, and the black start system comprises: an acquisition module used to acquire the closing overvoltage and multiple voltage influence parameters of the unloaded line during the past black start of the power grid; a weight calculation module used to calculate a first set of weights and a second set of weights for the closed-circuit overvoltage and the plurality of voltage influence parameters using a Pearson correlation coefficient and a cosine similarity; an error calculation module used to determine a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage influence parameters, the first weight set, and the second weight set using a recurrent neural network model; an optimization module used to update the first weight set and the second weight set according to the first error and the second error according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further to obtain a target weight set; and a prediction module that uses a recurrent neural network model to obtain a set of closed-circuit overvoltage target prediction values ​​based on the target weight set and multiple real-time voltage influence parameters in the current black start, and selects the line on which the minimum value in the set of closed-circuit overvoltage target prediction values ​​is located as the optimal line, which is used to supply power to the power grid to complete the black start of the power grid.

[0017] To achieve the above object, an embodiment of a third aspect of the present application proposes an electronic device, comprising: a processor; and a memory communicatively connected to the processor, wherein the memory stores computer-executable instructions, and the processor executes the computer-executable instructions stored in the memory to realize the method provided in the first aspect of the present application.

[0018] To achieve the above object, an embodiment of a fourth aspect of the present application proposes a computer-readable recording medium having computer-executable instructions stored thereon, the computer-executable instructions being used to implement the method provided in the first aspect of the present application when executed by a processor.

[0019] The present application provides a black start method, system, electronic device, and recording medium for a diesel combined combustion engine, which includes a diesel black start unit, which is used to start the gas generator through a starting device for the gas generator when the gas generator stops due to a loss of power in the power grid. The black start method includes the steps of obtaining a closed-circuit overvoltage and a plurality of voltage influence parameters of an unloaded line when the power grid has previously black started; calculating a first weight set and a second weight set for the closed-circuit overvoltage and the plurality of voltage influence parameters using a Pearson correlation coefficient and a cosine similarity; and calculating a first weight set and a second weight set for the closed-circuit overvoltage and the plurality of voltage influence parameters using a recurrent neural network model. determining a first error and a second error based on a number of voltage influence parameters, a first weight set, and a second weight set; updating the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set; and using a recurrent neural network model to obtain a closed-loop overvoltage target predicted value set based on the target weight set and the plurality of real-time voltage influence parameters in the current black start, and supplying power to the power grid to complete the black start of the power grid by selecting the line on which the minimum value in the closed-loop overvoltage target predicted value set is located as the optimal line. In this case, the influence of multiple voltage influence parameters on the closed circuit overvoltage is considered, and a first weight set and a second weight set are obtained using Pearson correlation coefficient and cosine similarity, and then a first error and a second error are obtained. A target weight set is obtained according to the first error and the second error after a predetermined number of iterations. A closed circuit overvoltage target prediction value set is obtained based on the target weight set and multiple real-time voltage influence parameters in the current black start, and then an optimal line is obtained. In this case, the target weight set fully considers the degree of influence of various factors on overvoltage and improves the accuracy of black start closed circuit overvoltage prediction, so that the closed circuit overvoltage of different starting lines can be better predicted, and thereby the optimal line can be selected.

[0020] Additional aspects and advantages of the present application will be set forth in part in the description that follows, and in part will be obvious from the description, or may be learned by practice of the present application. [Brief explanation of the drawings]

[0021] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the examples taken in conjunction with the drawings. [Figure 1] 1 is a schematic diagram illustrating the connection of a power plant to a power grid provided by an embodiment of the present application. [Figure 2] FIG. 1 is a schematic diagram showing the flow of a black start method for a diesel combined combustion engine provided by an embodiment of the present application. [Figure 3] FIG. 1 is a schematic diagram showing a specific flow of a black start method for a diesel combined combustion engine provided by an embodiment of the present application. [Figure 4] 1 is a block diagram of a black start system for a diesel combined combustion engine provided by an embodiment of the present application; DETAILED DESCRIPTION OF THE INVENTION

[0022]

[0023] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the drawings, where the same or similar numbers indicate the same or similar elements, or elements having the same or similar functions, throughout. The embodiments described below with reference to the drawings are illustrative and are used to explain the present application, and are not to be construed as limiting the present application.

[0023] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A black start method and system for a diesel combined combustion engine according to an embodiment of the present application will be described below with reference to the drawings.

[0024] The embodiments of the present application provide a black start method for a diesel combined combustion engine, which can better predict the closing overvoltage of different starting paths and select the optimal path.

[0025] In this application, a diesel black start unit is arranged in the power plant, and the diesel black start unit is used to start the gas generator via the gas generator starting equipment when the gas generator shuts down due to a loss of power in the power grid.

[0026] FIG. 1 is a schematic diagram illustrating the connection of a power plant to a power grid provided by an embodiment of the present application.

[0027] As shown in FIG. 1 , a power plant includes a diesel generator system (also referred to as a diesel black start unit) and a gas generator system. The diesel generator system is connected to the gas generator system, which is connected to a power grid via multiple paths. The gas generator system includes a gas generator starting device and a gas generator. Here, the gas generator is used for power generation. When the gas generator stops due to a loss of power in the power grid, the diesel black start unit supplies power to the gas generator starting device to restore a reference voltage value, starts the gas generator based on the reference voltage value, and then supplies power to the power grid via one of the multiple paths to complete the black start. To select the optimal path to complete the power grid black start, an overvoltage prediction system is provided to predict closed-circuit overvoltages of different paths. Here, the overvoltage prediction system is also referred to as a black start system for a diesel combined combustion engine. The overvoltage prediction system is used to perform the black start method for a diesel combined combustion engine according to the present application.

[0028] Figure 2 is a schematic diagram showing the flow of a black start method for a diesel combined combustion engine provided by an embodiment of the present application. Figure 3 is a schematic diagram showing a specific flow of a black start method for a diesel combined combustion engine provided by an embodiment of the present application.

[0029] As shown in FIG. 2, the black start method for the diesel combined combustion engine includes the following steps S101 to S105.

[0030] Step S101: Obtain the closed circuit overvoltage of the unloaded line and multiple voltage influence parameters when the power grid has experienced a black start in the past.

[0031] In step S101, the number of closed-circuit overvoltages of unloaded lines acquired in the past when the power grid black started can be n. All acquired closed-circuit overvoltages are also called a closed-circuit overvoltage set. A closed-circuit overvoltage set consisting of n closed-circuit overvoltages can be represented by the symbol Y.

[0032] In step S101, each closed circuit overvoltage corresponds to a plurality of voltage-influencing parameters, including the length of the closed line, the line shunt reactor compensation value, the power supply resistance, the power supply leakage reactance, the line resistance per kilometer, the line positive-sequence reactance per kilometer, and the closed circuit initial phase angle. For example, the voltage-influencing parameters during a past black start can be represented by historical raw data X, which can be expressed as X=(X1, X2, X3, X4, X5, X6, X7), where X1 is the length of n closed lines, X2 is the line shunt reactor compensation value, X3 is the power supply resistance, X4 is the line leakage reactance, X5 is the line resistance per kilometer, X6 is the line positive-sequence reactance per kilometer, and X7 is the closed circuit initial phase angle.

[0033] Step S102: Calculate a first weight set and a second weight set for the closed circuit overvoltage and the multiple voltage influence parameters using Pearson correlation coefficient and cosine similarity.

[0034] In step S102, calculating a first weight set and a second weight set for the closed-circuit overvoltage and the multiple voltage-influence parameters using the Pearson correlation coefficient and the cosine similarity includes: calculating a first correlation coefficient set for the closed-circuit overvoltage and the multiple voltage-influence parameters using the Pearson correlation coefficient; obtaining the first weight set based on the sum of the absolute value of each first correlation coefficient in the first correlation coefficient set and the absolute values ​​of all the first correlation coefficients; calculating a second correlation coefficient set for the closed-circuit overvoltage and the multiple voltage-influence parameters using the cosine similarity; and obtaining the second weight set based on the sum of the absolute value of each second correlation coefficient in the second correlation coefficient set and the absolute values ​​of all the second correlation coefficients.

[0035] In step S102, as can be easily understood, the Pearson correlation coefficient can measure the presence or absence of a linear correlation between two feature quantities and the magnitude of the correlation. Therefore, in step S102, the Pearson correlation coefficient is used to measure the correlation between the black start overvoltage (i.e., the closed circuit overvoltage) and other feature quantities (i.e., multiple voltage influence parameters), and the Pearson correlation coefficient satisfies the following: TIFF2025126101000002.tif19170 where, TIFF2025126101000003.tif8170i shows the strength of correlation between the voltage influence parameter and the closed circuit overvoltage, TIFF2025126101000004.tif8170A positive value indicates a positive correlation between the i-th voltage influence parameter and the closed circuit overvoltage. TIFF2025126101000005.tif8170A negative value indicates a negative correlation between the i-th voltage influence parameter and the closed circuit overvoltage, i=1, 2, ..., 7, i.e., TIFF2025126101000006.tif10170TIFF2025126101000007.tif8170Can be considered as a 7x1 matrix. TIFF2025126101000008.tif8170 denotes the sum of n values ​​of the ith voltage influence parameter in the raw dataset of influence parameters, TIFF2025126101000009.tif8170 shows the sum of n closed circuit overvoltages. E() indicates covariance.

[0036] Any first weight in the first weight set satisfies the following: In the TIFF2025126101000010.tif20170 format, TIFF2025126101000011.tif10170Indicates the first weight of the i-th voltage influence parameter, where i=1, 2, …, 7.

[0037] In step S102, as can be easily understood, the cosine similarity measures whether there is a linear correlation between two features and the magnitude of the correlation by calculating the cosine value of the angle between two vectors. Therefore, in step S102, the cosine similarity is used to measure the correlation between the black start overvoltage and other features, and the cosine similarity satisfies the following: In the TIFF2025126101000012.tif14170 format, TIFF2025126101000013.tif8170This shows the strength of correlation between the i-th voltage influence parameter and the closed circuit overvoltage, where i=1, 2, ..., 7, i.e., TIFF2025126101000014.tif10170TIFF2025126101000015.tif8170It can be considered as a 7x1 matrix, where TIFF2025126101000016.tif9170 Shows the cosine function.

[0038] Any second weight in the second weight set satisfies the following: In the TIFF2025126101000017.tif21170 format, TIFF2025126101000018.tif10170Indicates the second weight of the i-th voltage influence parameter, where i=1, 2, …, 7.

[0039] Step S103: Using a recurrent neural network model, determine a first error and a second error based on the closed circuit overvoltage, the plurality of voltage influence parameters, the first weight set and the second weight set.

[0040] In step S103, determining a first error and a second error based on the closed circuit overvoltage, the plurality of voltage-influencing parameters, the first weight set, and the second weight set using the recurrent neural network model includes: obtaining first model input data based on the plurality of voltage-influencing parameters and the first weight set, inputting the first model input data into the recurrent neural network model to obtain a first set of overvoltage prediction values, and obtaining a first error based on the closed circuit overvoltage and the first set of overvoltage prediction values; obtaining second model input data based on the plurality of voltage-influencing parameters and the second weight set, inputting the second model input data into the recurrent neural network model to obtain a second set of overvoltage prediction values, and obtaining a second error based on the closed circuit overvoltage and the second set of overvoltage prediction values.

[0041] As can be easily understood, in step S103, the recurrent neural network model can be selected as a GRU (gated recurrent unit) network model. The GRU network is an improved model of the LSTM (long short-term memory network) network, which integrates the forget gate and the input gate into the update gate, thereby reducing the learning parameters of the network to a certain extent and ensuring the storage of effective information.

[0042] In step S103, the first model input data can be represented by X', the first set of overvoltage prediction values ​​can be represented by Y', the second model input data can be represented by X'', and the second set of overvoltage prediction values ​​can be represented by Y''. Specifically, the first model input data X' can be represented by TIFF2025126101000019.tif10170, and the second model input data X" is TIFF2025126101000020.tif9170, where X indicates historical raw data, TIFF2025126101000021.tif10170 indicates the first weight of the i-th voltage influence parameter, TIFF2025126101000022.tif10170 shows the second weight of the i-th voltage influence parameter. By substituting X' and X" into the GRU network model and making predictions, we obtain the first overvoltage prediction value set Y' and the second overvoltage prediction value set Y" (see Figure 3).

[0043] Specifically, the first error and the second error can be calculated using the average relative error, where: TIFF2025126101000023.tif9170 or less. TIFF2025126101000024.tif17170 where, TIFF2025126101000025.tif8170 actual value, and the actual value is the i-th value in the closed circuit overvoltage set Y; TIFF2025126101000026.tif9170If it is the i-th value in the first overvoltage prediction value set Y', TIFF2025126101000027.tif10170If it is the i-th value in the second overvoltage prediction value set Y”, TIFF2025126101000028.tif10170N is the total number of samples, i.e., the number of closed-circuit overvoltages in the closed-circuit overvoltage set Y.

[0044] Step S104: based on the first error and the second error, update the first weight set and the second weight set according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further obtain a target weight set.

[0045] In step S104, updating the first weight set and the second weight set according to the first error and the second error according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set, The method includes determining whether the current number of iterations is equal to the predetermined number of iterations, and if not, comparing the first error and the second error, updating the first weight set and the second weight set according to different requirements based on the comparison result until the current number of iterations is equal to the predetermined number of iterations, and obtaining a first weight optimization set and a second weight optimization set based on the first weight set and the second weight set under each iteration; determining a first error target value and a second error target value based on the closed-circuit overvoltage, the plurality of voltage influence parameters, the first weight optimization set and the second weight optimization set using a recurrent neural network model; and selecting the weight optimization set corresponding to the smallest value of the first error target value and the second error target value as the target weight set.

[0046] Here, updating the first weight set and the second weight set according to different requirements based on the comparison result until the current number of iterations is equal to the predetermined number of iterations includes: if the first error is greater than the second error, updating the first weight set and the second weight set according to a first step length and a second step length, respectively; and if the first error is equal to or less than the second error, updating the first weight set and the second weight set according to a first proportion and a second proportion, respectively.

[0047] Obtaining the first and second weight optimization sets based on the first and second weight sets under each number of iterations includes obtaining the first and second weight optimization sets based on the first and second weight sets in the final iteration and the first and second weight sets under the number of iterations corresponding to the case where the sum of the first error and the second error is smallest, where, for example, when the predetermined number of iterations is 20, any first weight optimization value in the first weight optimization set satisfies the following: In the TIFF2025126101000029.tif9170 format, TIFF2025126101000030.tif10170 is the first weight optimization value of the i-th voltage influence parameter, TIFF2025126101000031.tif1017020 is the first weight of the i-th voltage influence parameter at the iteration count, TIFF2025126101000032.tif1217020 is the second weight of the i-th voltage influence parameter at iteration count.

[0048] Any second weight optimization value in the second weight optimization set satisfies the following:

[0049] In the TIFF2025126101000033.tif10170 format, TIFF2025126101000034.tif10170 is the second weight optimization value of the i-th voltage influence parameter, TIFF2025126101000035.tif11170 is the first weight of the i-th voltage influence parameter under the number of iterations corresponding to the smallest case, TIFF2025126101000036.tif10170The second weight set for the i-th voltage influence parameter under the number of iterations corresponding to the minimum case.

[0050] Specifically, as shown in FIG. 3, after obtaining the first error and the second error, it is determined whether the current number of repetitions T (the initial value of the number of repetitions is 1) is equal to a predetermined number of repetitions (for example, 20 times). If not, TIFF2025126101000037.tif10170 Update the first weight set and the second weight set according to the first step length (e.g., +0.05) and the second step length (e.g., -0.05), respectively, i.e., TIFF2025126101000038.tif10170Further update the number of iterations (T=T+1), go back and re-obtain the first error and the second error based on the updated first weight set and second weight set, and terminate the update if the current number of iterations is equal to the predetermined number of iterations; if the current number of iterations is not equal to the predetermined number of iterations, continue to determine the new first error and the new second error; TIFF2025126101000039.tif10170 Update the first weight set and the second weight set according to the first proportionality (e.g., 0.99) and the second proportionality (e.g., 1.01), respectively, i.e., TIFF2025126101000040.tif9170Further update the number of iterations (T=T+1), go back and re-obtain the first error and second error based on the updated first and second weight sets, and terminate the update and move away from recurrent until the current number of iterations is equal to the specified number of iterations.

[0051] As shown in FIG. 3, a first weight optimization set and a second weight optimization set are calculated, and corresponding model input data are obtained according to the first weight optimization set and the second weight optimization set. The first weight optimization set corresponds to the model input data X'(2), and the model input data X'(2) is TIFF2025126101000041.tif10170The second weight optimization set corresponds to the model input data X''(2), and the model input data X''(2) is TIFF2025126101000042.tif12170, where X represents the historical raw data. Substituting X'(2) and X''(2) into the GRU network model for prediction, respectively, obtains the corresponding set of overvoltage prediction values. Calculate the corresponding first and second error target values ​​by referring to the above average relative error together with the closed-circuit overvoltage. The smallest weight coefficient (i.e., the weight optimization set) of the first and second error target values ​​is determined. TIFF2025126101000043.tif8170 (i.e., the target weight set).

[0052] Step S105: Use the recurrent neural network model to obtain a set of closed-loop overvoltage target prediction values ​​based on the target weight set and the multiple real-time voltage influence parameters in the current black start, and select the line with the minimum value in the set of closed-loop overvoltage target prediction values ​​as the optimal line to supply power to the power grid to complete the black start of the power grid.

[0053] In step S105, the multiple real-time voltage influence parameters may use a real-time data set X0, which includes the length of the closed line, the line shunt reactor compensation value, the power supply resistance, the power supply leakage reactance, the line resistance per kilometer, the line positive-sequence reactance per kilometer, and the closed-line initial phase angle at the current black start, all of which are acquired in real time.

[0054] Based on TIFF2025126101000044.tif9170 and the real-time dataset X0, the target model input data M is obtained, namely, TIFF2025126101000045.tif7170The target model input data M is substituted into the GRU network model for prediction, and the final prediction result (i.e., a set of closed-loop overvoltage target prediction values) is obtained.From the set of closed-loop overvoltage target prediction values, the line on which the smallest closed-loop overvoltage target prediction value is located is selected as the optimal line, and power is supplied to the power grid to complete the black start of the power grid.

[0055] To verify the effectiveness of the method of the present application, an equivalent network is constructed to calculate the closed-circuit overvoltage of a 500 kV unloaded line during black start. In the following, this network is used as an operation example to verify the effectiveness of the fast prediction method for stating overvoltages based on the unloaded line during black start of the combined model. Set the power supply resistance Rs = 15 to 45 Ω in 10 Ω intervals, the line length l = 280 to 350 km in 40 km intervals, the shunt reactor compensation Q = 70 to 80 MVAR in 5 MVAR intervals, and the power supply leakage reactance xs = 125 to 140 Ω in 5 Ω intervals.

[0056] The multiple voltage influence parameters include seven voltage influence parameters: length of the closed line, line shunt reactor compensation value, source resistance, source leakage reactance, line resistance per km, line positive-sequence reactance per km, and closed line initial phase angle.

[0057] The model evaluation criteria are selected as mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE), and all models use MAE as the loss function during training. During the experiments, comparisons are also made with the independent Pearson coefficient-GRU model and cosine similarity-GRU model. The results are shown in Table 1.

[0058] As can be seen from the table TIFF2025126101000046.tif43170, the prediction accuracy using the model of the present application is higher than the prediction accuracy using only the Pearson coefficient or cosine similarity, demonstrating the importance of the inertia weighting coefficient for improving accuracy.

[0059] To realize the above embodiment, the present application further proposes a black start system for a diesel combined combustion engine, in which a diesel black start unit is arranged, and the diesel black start unit is used to start the gas generator through the gas generator starting equipment when the gas generator stops due to power loss in the power grid.

[0060] FIG. 4 is a block diagram of a black start system for a diesel combined combustion engine provided by an embodiment of the present application.

[0061] As shown in FIG. 4, the black start system for the diesel combined combustion engine includes an acquisition module 11, a weight calculation module 12, an error calculation module 13, an optimization module 14 and a prediction module 15, wherein: The acquisition module 11 is used to acquire the closed circuit overvoltage and multiple voltage influence parameters of the unloaded line when the power grid has experienced a black start in the past; The weight calculation module 12 is used to calculate a first weight set and a second weight set for the closed circuit overvoltage and the plurality of voltage influence parameters using Pearson correlation coefficient and cosine similarity; the error calculation module 13 is used to determine a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage influence parameters, the first weight set and the second weight set using a recurrent neural network model; the optimization module 14 is used to update the first weight set and the second weight set according to the first error and the second error according to a predetermined number of iterations to obtain a first weight optimization set and a second weight optimization set, and further to obtain a target weight set; The prediction module 15 uses a recurrent neural network model to obtain a set of closed-circuit overvoltage target prediction values ​​based on a set of target weights and a plurality of real-time voltage influence parameters in the current black start, and selects the line on which the minimum value in the set of closed-circuit overvoltage target prediction values ​​is located as the optimal line, which is used to supply power to the power grid to complete the black start of the power grid.

[0062] Furthermore, in a possible implementation form of the embodiments of the present application, the weight calculation module 12 is specifically used for calculating a first set of correlation coefficients between the closed-circuit overvoltage and the multiple voltage influence parameters using Pearson correlation coefficients, and obtaining a first set of weights based on the sum of the absolute value of each first correlation coefficient in the first correlation coefficient set and the absolute values ​​of all the first correlation coefficients, and for calculating a second set of correlation coefficients between the closed-circuit overvoltage and the multiple voltage influence parameters using cosine similarity, and obtaining a second set of weights based on the sum of the absolute value of each second correlation coefficient in the second correlation coefficient set and the absolute values ​​of all the second correlation coefficients.

[0063] Furthermore, in a possible implementation form of the embodiments of the present application, the error calculation module 13 is specifically used for: obtaining first model input data based on the plurality of voltage-influencing parameters and a first set of weights; inputting the first model input data into a recurrent neural network model to obtain a first set of overvoltage prediction values; and obtaining a first error based on the closed-circuit overvoltage and the first set of overvoltage prediction values; and obtaining second model input data based on the plurality of voltage-influencing parameters and a second set of weights; inputting the second model input data into the recurrent neural network model to obtain a second set of overvoltage prediction values; and obtaining a second error based on the closed-circuit overvoltage and the second set of overvoltage prediction values.

[0064] Furthermore, in a possible implementation form of the embodiments of the present application, the optimization module 14 is specifically used for: determining whether the current iteration number is equal to the predetermined iteration number; if not, comparing the first error and the second error; updating the first weight set and the second weight set according to different requirements based on the comparison result until the current iteration number is equal to the predetermined iteration number; obtaining a first weight optimization set and a second weight optimization set based on the first weight set and the second weight set under each iteration number; determining a first error target value and a second error target value based on the closed-circuit overvoltage, the multiple voltage influence parameters, the first weight optimization set and the second weight optimization set using a recurrent neural network model; and selecting the weight optimization set corresponding to the smallest value of the first error target value and the second error target value as the target weight set.

[0065] Furthermore, in a possible implementation form of the embodiments of the present application, in the optimization module 14, updating the first weight set and the second weight set according to different requirements based on the comparison result until the current iteration number is equal to a predetermined iteration number includes: if the first error is greater than the second error, updating the first weight set and the second weight set according to a first step length and a second step length, respectively; and if the first error is less than or equal to the second error, updating the first weight set and the second weight set according to a first proportion and a second proportion, respectively.

[0066] Furthermore, in a possible implementation form of the embodiments of the present application, in the optimization module 14, obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set under each iteration number includes obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set in the last iteration and the first weight set and the second weight set under the iteration number corresponding to the case where the sum of the first error and the second error is smallest.

[0067] The above description of the embodiment of the black start method for a diesel combined combustion engine is also applicable to the black start system for a diesel combined combustion engine according to the embodiment, and will not be repeated here.

[0068] In an embodiment of the present application, a diesel black start unit is provided, and the diesel black start unit is used to start the gas generator through the gas generator starting equipment when the gas generator is shut down due to a power loss in the power grid. The black start method includes the steps of: obtaining a closed-circuit overvoltage and a plurality of voltage influence parameters of an unloaded line when the power grid has previously black started; calculating a first weight set and a second weight set for the closed-circuit overvoltage and the plurality of voltage influence parameters using a Pearson correlation coefficient and a cosine similarity; and calculating a first weight set and a second weight set for the closed-circuit overvoltage, the plurality of voltage influence parameters, the first weight set and the second weight set using a recurrent neural network model. and a second weight set; updating the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set; and using a recurrent neural network model to obtain a closed-loop overvoltage target prediction value set based on the target weight set and a plurality of real-time voltage influence parameters in the current black start, and supplying power to the power grid to complete the black start of the power grid by selecting the line on which the minimum value in the closed-loop overvoltage target prediction value set is located as the optimal line. In this case, the influence of multiple voltage influence parameters on the closed circuit overvoltage is considered, and a first weight set and a second weight set are obtained using Pearson correlation coefficient and cosine similarity, and then a first error and a second error are obtained. A target weight set is obtained according to the first error and the second error through a predetermined number of iterations. A closed circuit overvoltage target prediction value set is obtained based on the target weight set and the multiple real-time voltage influence parameters in the current black start, and then an optimal line is obtained. In this case, the target weight set fully takes into account the degree of influence of various factors on overvoltage and improves the accuracy of the closed circuit overvoltage prediction for the black start, so that the closed circuit overvoltage of different starting paths can be better predicted, and thereby the optimal path (i.e., the optimal line) can be selected.

[0069] The method and system of the present application take into account the differences in the degree to which different factors affect overvoltages and further quantitatively analyze the different factors, thereby making it possible to better predict the no-load line closing overvoltages of different starting paths and select the optimal path.

[0070] To realize the above-mentioned embodiments, the present application further proposes an electronic device, including a processor and a memory communicatively connected to the processor, wherein the memory stores computer-executable instructions, and the processor realizes the method provided by the aforementioned embodiments by executing the computer-executable instructions stored in the memory.

[0071] To realize the above embodiments, the present application further proposes a computer-readable recording medium, in which computer-executable instructions are stored, which, when executed by a processor, are used to realize the methods provided by the aforementioned embodiments.

[0072] To realize the above embodiments, the present application further proposes a computer program product, which includes a computer program, which, when executed by a processor, realizes the method provided by the aforementioned embodiments.

[0073] In the descriptions of the above embodiments, terms such as "one embodiment," "some embodiments," "example," "particular example," or "some examples" mean that the specific feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, those skilled in the art may combine different embodiments or examples, and features of different embodiments or examples, described herein without mutual contradiction.

[0074] It should be noted that the terms "first" and "second" are for descriptive purposes only and should not be understood as indicating or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature qualified by "first" or "second" may explicitly or implicitly include at least one of the feature. In the description of this application, "plurality" means at least two, e.g., two, three, etc., unless explicitly and specifically limited.

[0075] Any process or method description shown in a flowchart or otherwise described herein may be understood to represent a module, fragment, or portion comprising one or more executable instruction codes for implementing customized logical functions or process steps, and as will be understood by those skilled in the art, the scope of the preferred embodiments of the present application includes additional implementations that perform functions in a substantially concurrent manner or in reverse order, depending on the functionality involved, rather than in the order shown or discussed.

[0076] The logic and / or steps illustrated in the flowcharts or described elsewhere herein can be viewed, for example, as a sequential listing of executable instructions that implement logical functions, and specifically can be embodied on any computer-readable medium for use by or in combination with an instruction execution system, device, or apparatus (e.g., a computer-based system, a system including a processor, or other system that retrieves and executes instructions from another instruction execution system, device, or apparatus). As used herein, the term "computer-readable medium" refers to any device that can store, preserve, communicate, transmit, or transmit a program for use by or with an instruction execution system, device, or apparatus. More specific examples (non-exhaustive list) of computer-readable media include an electrical connection having one or more wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CD-ROM). The computer readable medium may also be paper or other suitable medium on which the program may be printed, and the program may be obtained electronically and stored in computer memory, for example by optically scanning the paper or other medium and then editing, decrypting, or otherwise processing as needed.

[0077] As will be understood, each part of the present application may be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods may be implemented by software or firmware stored in a memory and executed by an appropriate instruction execution system. For example, when implemented by hardware as in other embodiments, they may be implemented by any one or combination of the following technologies known in the art: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, special purpose integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0078] Those skilled in the art will understand that all or part of the steps performed to realize the method of the above embodiments can be achieved by a program instructing relevant hardware, which may be stored in a computer-readable recording medium, and when executed, constitutes one or a combination of the steps of the method embodiments.

[0079] Furthermore, each functional unit in various embodiments of the present application may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The integrated module may be realized in the form of hardware or a software functional module. When the integrated module is implemented as a software functional module and sold or used as a separate product, it may be stored in a computer-readable storage medium.

[0080] The storage medium may be a read-only memory, a magnetic disk, an optical disk, etc. Although the above has shown and described embodiments of the present application, it should be understood that the above embodiments are illustrative and should not be construed as limitations on the present application, and that those skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A diesel black start unit is provided, and the diesel black start unit is used to start the gas generator through the gas generator starting equipment when the gas generator stops due to a loss of power in the power grid, and the black start method includes: obtaining a closing overvoltage and a plurality of voltage influence parameters of an unloaded line when the power grid has previously experienced a black start; calculating a first set of weights and a second set of weights for the closed-circuit overvoltage and the plurality of voltage-influence parameters using Pearson correlation coefficient and cosine similarity; determining a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage-influencing parameters, the first set of weights, and the second set of weights using a recurrent neural network model; updating the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain a first weight optimization set and a second weight optimization set, and further obtaining a target weight set; using a recurrent neural network model to obtain a set of closed-circuit overvoltage target prediction values ​​based on the target weight set and a plurality of real-time voltage influence parameters in the current black start; and supplying power to the power grid by selecting the line on which the minimum value in the set of closed-circuit overvoltage target prediction values ​​is located as the optimal line to complete the black start of the power grid.

2. calculating a first set of weights and a second set of weights for the closed-circuit overvoltage and the plurality of voltage-influence parameters using the Pearson correlation coefficient and the cosine similarity described above, calculating a first set of correlation coefficients for the closed-circuit overvoltage and the plurality of voltage-influence parameters using a Pearson correlation coefficient; and obtaining a first set of weights based on the sum of the absolute value of each first correlation coefficient and the absolute values ​​of all the first correlation coefficients in the first correlation coefficient set; 2. The black-start method for a diesel combined combustion engine according to claim 1, further comprising: calculating a second set of correlation coefficients for the closed-circuit overvoltage and the plurality of voltage-influencing parameters using cosine similarity; and obtaining a second set of weights based on a sum of an absolute value of each second correlation coefficient and an absolute value of all the second correlation coefficients in the second correlation coefficient set.

3. Determining a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage-influencing parameters, the first weight set, and the second weight set using the recurrent neural network model described above includes: obtaining first model input data based on the plurality of voltage-influencing parameters and the first set of weights, inputting the first model input data into a recurrent neural network model to obtain a first set of overvoltage prediction values, and obtaining a first error based on the closed-circuit overvoltage and the first set of overvoltage prediction values; 2. The black-start method for a diesel combined combustion engine according to claim 1, further comprising: obtaining second model input data based on the plurality of voltage-influencing parameters and the second set of weights; inputting the second model input data into a recurrent neural network model to obtain a second set of overvoltage prediction values; and obtaining a second error based on the closed-circuit overvoltage and the second set of overvoltage prediction values.

4. The method of updating the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain the first weight optimization set and the second weight optimization set, and further obtaining the target weight set, includes:

2. The black start method for a diesel combined combustion engine according to claim 1, further comprising: determining whether the current number of iterations is equal to the predetermined number of iterations; if not, comparing the first error with the second error; updating the first weight set and the second weight set according to different requirements based on the comparison result until the current number of iterations is equal to the predetermined number of iterations; and obtaining the first weight optimization set and the second weight optimization set according to the first weight set and the second weight set under each iteration number.

5. The method of updating the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain the first weight optimization set and the second weight optimization set, and further obtaining the target weight set, includes: determining a first error target value and a second error target value based on the closed-circuit overvoltage, the plurality of voltage-influencing parameters, the first weight optimization set, and the second weight optimization set using a recurrent neural network model; 5. The black start method for a diesel combined combustion engine according to claim 4, further comprising: selecting a weight optimization set corresponding to the smallest value among the first error target value and the second error target value as the target weight set.

6. Updating the first weight set and the second weight set according to different requirements based on the comparison result until the current iteration number is equal to the predetermined iteration number, If the first error is greater than the second error, updating the first weight set and the second weight set according to the first step length and the second step length, respectively; and updating the first weight set and the second weight set according to the first proportionality and the second proportionality, respectively, if the first error is less than or equal to the second error.

7. Obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set under each of the above-mentioned iteration numbers includes:

5. The black start method for a diesel combined combustion engine according to claim 4, further comprising obtaining the first weight optimization set and the second weight optimization set based on the first weight set and the second weight set in the final iteration and the first weight set and the second weight set under the iteration number corresponding to the case where the sum of the first error and the second error is minimum.

8. A diesel black start unit is arranged, and the diesel black start unit is used to start the gas generator through the gas generator starting equipment when the gas generator stops due to a loss of power in the power grid, and the black start system includes: an acquisition module used to acquire the closing overvoltage and multiple voltage influence parameters of the unloaded line during the past black start of the power grid; a weight calculation module used to calculate a first set of weights and a second set of weights for the closed-circuit overvoltage and the plurality of voltage-influence parameters using a Pearson correlation coefficient and a cosine similarity; an error calculation module used to determine a first error and a second error based on the closed-circuit overvoltage, the plurality of voltage-influencing parameters, the first weight set, and the second weight set using a recurrent neural network model; an optimization module used to update the first weight set and the second weight set according to a predetermined number of iterations based on the first error and the second error to obtain a first weight optimization set and a second weight optimization set, and further to obtain a target weight set; and a prediction module that uses a recurrent neural network model to obtain a set of closed-circuit overvoltage target prediction values ​​based on the target weight set and a plurality of real-time voltage influence parameters in the current black start, and that selects a line on which the minimum value in the set of closed-circuit overvoltage target prediction values ​​is located as an optimal line, and supplies power to the power grid to complete the black start of the power grid.

9. a processor and a memory communicatively connected to the processor; The memory stores computer executable instructions; The electronic device, characterized in that the processor implements the method according to any one of claims 1 to 7 by executing computer-executable instructions stored in the memory.

10. A computer-readable recording medium having stored thereon computer-executable instructions, the computer-executable instructions being used to implement the method of any one of claims 1 to 7 when executed by a processor.

Images