Power power supply quantity demand prediction method, storage medium and storage terminal

Abstract

A power supply quantity demand prediction method, a storage medium and a storage terminal, the method comprising: providing a plurality of data pairs, the plurality of data pairs comprising a plurality of feature parameters affecting power supply quantity and corresponding power supply quantity; obtaining an initial neural network prediction model based on the plurality of data pairs; the initial neural network prediction model is trained and tested, a neural network prediction model is obtained, and the error of the neural network prediction model is within a preset range; and predicting the demand of the dynamic power supply quantity according to the neural network prediction model. The power supply quantity demand prediction method can accurately predict the power supply quantity.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a method, a storage medium and a storage terminal for predicting power supply demand. Background Art

[0002] Electricity consumption forecasting plays a vital role in the operation of power systems. Accurate electricity consumption forecasting can improve the safety and reliability of power grid operation. At present, enterprises mainly judge and arrange the number of generators, transformers, uninterruptible power supplies (UPS) and busbars based on experience, which will lead to waste and excessive consumption of electricity. Summary of the invention

[0003] The technical problem solved by the present invention is to provide a method, a storage medium and a storage terminal for predicting power supply demand, so as to improve the accuracy of power prediction.

[0004] In order to solve the above technical problems, the technical solution of the present invention provides a method for predicting power supply demand, including: providing a number of data pairs, the data pairs including a number of characteristic parameters affecting the power supply and the corresponding power supply; obtaining an initial neural network prediction model based on the data pairs; training and testing the initial neural network prediction model to obtain a neural network prediction model, the error of the neural network prediction model being within a preset range; and predicting the power supply demand according to the neural network prediction model.

[0005] Optionally, the initial neural network prediction model includes an input layer, a hidden layer and an output layer, and the method for obtaining the initial neural network prediction model based on several of the data pairs includes: using the several characteristic parameters affecting the power supply as neurons of the input layer; obtaining the number of nodes of the input layer according to the number of characteristic parameters affecting the power supply; using the power supply as neurons of the output layer, and obtaining the number of nodes of the output layer according to the power supply; obtaining the number of nodes of the hidden layer according to the number of nodes of the input layer and the number of nodes of the output layer; obtaining the initial neural network prediction model according to the neurons of the input layer, the number of nodes of the input layer, the neurons of the output layer, the number of nodes of the output layer, and the number of nodes of the hidden layer.

[0006] Optionally, several characteristic parameters that affect the power supply include: several sets of transformers with different voltages, several sets of uninterruptible power supplies with different voltages, several busbars with different rated currents, and several generators.

[0007] Optionally, several sets of transformers with different voltages include: several sets of transformers of 208V series, several sets of transformers of 308V series and several sets of transformers of 480V series; several sets of uninterruptible power supplies with different voltages include: several sets of UPS208V series, several sets of UPS380V series and several sets of UPS480V series; several busbars with different rated currents include: several busbars of 1600A series, several busbars of 2000A series and several busbars of 2500A series; several sets of the generators include: several 2200KW generators.

[0008] Optionally, the number of nodes in the input layer is obtained according to the number of characteristic parameters affecting the power supply, and the number of nodes in the input layer is 10; the number of nodes in the output layer is obtained according to the power supply, and the number of nodes in the output layer is 1.

[0009] Optionally, the number of nodes in the hidden layer is obtained according to the number of nodes in the input layer and the number of nodes in the output layer, and the number of nodes in the hidden layer is Among them, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

[0010] Optionally, before training and testing the initial neural network prediction model, it also includes: obtaining connection weights based on the number of nodes in the input layer, the number of nodes in the output layer, and the number of nodes in the hidden layer; obtaining the hidden layer activation function, the output layer activation function, the network training function, and the network performance function.

[0011] Optionally, a connection weight is obtained according to the number of nodes in the input layer, the number of nodes in the output layer and the number of nodes in the hidden layer, and the connection weight w=m×n+n×h+n+h, wherein w is the connection weight, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

[0012] Optionally, the hidden layer excitation function includes an S function; the output layer excitation function includes a linear function; the network training function includes a momentum BP algorithm with a variable learning rate; and the network performance function includes a mean square error function.

[0013] Optionally, the initial neural network prediction model is trained and tested, including: providing a sample set including several data pairs; processing and grouping the sample set to obtain training samples, verification samples and test samples; presetting the number of iterations, expected error and learning rate, and training the initial neural network prediction model according to the training samples; optimizing the initial neural network prediction model according to the verification samples; predicting the initial neural network prediction model according to the test samples until the initial neural network prediction model converges, and obtaining a neural network prediction model, wherein the expected error of the neural network prediction model is within a preset range.

[0014] Optionally, the demand for power supply is predicted according to the neural network prediction model, including: based on the neural network prediction model, according to the demand for power supply, obtaining the number of several sets of transformers with different voltages, the number of several sets of uninterruptible power supplies with different voltages, the number of busbars with different rated currents, and the number of generators.

[0015] Correspondingly, the technical solution of the present invention also provides a storage medium on which computer instructions are stored, and the steps of the above method are executed when the computer instructions are executed.

[0016] Correspondingly, the technical solution of the present invention also provides a storage terminal, including a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor, and the processor executes the steps of the above method when executing the computer instructions.

[0017] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0018] The method for predicting power supply demand of the present invention obtains a neural network prediction model based on several data pairs consisting of several characteristic parameters affecting the power supply and the corresponding power supply. The neural network prediction model can accurately predict the power supply, improve economic benefits and ensure the safe operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figures 1 to 3 is a flow chart of a method for predicting power supply demand in an embodiment of the present invention;

[0020] Figure 4 Schematic diagram of an initial neural network prediction model in an embodiment of the present invention;

[0021] Figure 5 It is a partial data schematic diagram of several characteristic parameters in an embodiment of the present invention;

[0022] Figure 6 is a graph showing the performance and error of a neural network in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned objects, features and beneficial effects of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0024] Figures 1 to 3 It is a flow chart of the method for predicting power supply demand in an embodiment of the present invention.

[0025] Please refer to Figure 1 , the method for predicting power supply demand includes:

[0026] Step S10: providing a plurality of data pairs, wherein the plurality of data pairs include a plurality of characteristic parameters affecting the power supply amount and the corresponding power supply amount;

[0027] Step S20: obtaining an initial neural network prediction model based on the plurality of data pairs;

[0028] Step S30: training and testing the initial neural network prediction model to obtain a neural network prediction model, wherein the error of the neural network prediction model is within a preset range;

[0029] Step S40: predicting the demand for power supply according to the neural network prediction model.

[0030] The method for predicting power supply demand obtains a neural network prediction model based on a number of data pairs consisting of a number of characteristic parameters affecting the power supply and the corresponding power supply. The neural network prediction model can accurately predict the power supply, improve economic benefits and ensure the safe operation of the power system.

[0031] Next, each step is analyzed and explained.

[0032] Please continue to refer to Figure 1 , executing step S10: providing a plurality of data pairs, wherein the plurality of data pairs include a plurality of characteristic parameters affecting the power supply amount and the corresponding power supply amount.

[0033] In this embodiment, several characteristic parameters affecting the power supply include: several sets of transformers with different voltages, several sets of uninterruptible power supplies (UPS) with different voltages, several buses with different rated currents, and several generators.

[0034] In this embodiment, several sets of transformers with different voltages include: several sets of transformers of 208V series, several sets of transformers of 308V series and several sets of transformers of 480V series. Among them, 208V, 308V and 480V are rated voltages of transformers. In this embodiment, only three transformer series with different rated voltages are schematically provided. In other embodiments, the rated voltage of the transformer can also be adjusted according to actual needs.

[0035] In this embodiment, the number of transformers of the 208V series is several sets, the number of transformers of the 308V series is several sets, and the number of transformers of the 480V series is several sets.

[0036] In this embodiment, several sets of uninterruptible power supplies with different voltages include: several sets of UPS208V series, several sets of UPS380V series and several sets of UPS480V series. Among them, 208V, 380V and 480V are the rated voltages of the uninterruptible power supplies. In this embodiment, only three uninterruptible power supply series with different rated voltages are schematically provided. In other embodiments, the rated voltage of the uninterruptible power supply can also be adjusted according to actual needs.

[0037] In this embodiment, the number of UPS208V series is several sets, the number of UPS380V series is several sets, and the number of UPS480V series is several sets.

[0038] In this embodiment, the plurality of busbars with different rated currents include: a plurality of busbars of 1600A series, a plurality of busbars of 2000A series, and a plurality of busbars of 2500A series. Among them, 1600A, 2000A, and 2500A are the rated currents of the busbars, and only three busbar series with different rated currents are schematically provided in this embodiment. In other embodiments, the rated current of the busbar can also be adjusted according to actual needs.

[0039] In this embodiment, the number of busbars 1600A series is several, the number of busbars 2000A series is several, and the number of busbars 2500A is several.

[0040] In this embodiment, the plurality of generators include: a plurality of 2200KW generators. 2200KW is the rated power of the generator, and only one rated power generator is schematically provided in this embodiment. In other embodiments, the rated power and power type of the generator can also be adjusted according to actual needs.

[0041] In this embodiment, the data pairs include: several sets of transformer 208V series, several sets of transformer 308V series, several sets of transformer 480V series, several sets of UPS 208V series, several sets of UPS 380V series, several sets of UPS 480V series, several bus 1600A series, several bus 2000A series, several bus 2500A series and several 2200KW generators. As the number of transformers, the number of uninterruptible power supplies, the number of busbars and the number of generators change, the corresponding power supply also changes.

[0042] Please continue to refer to Figure 1 , execute step S20: obtain an initial neural network prediction model based on the plurality of data pairs.

[0043] The initial neural network prediction model includes an input layer, a hidden layer and an output layer.

[0044] Figure 4 This is a schematic diagram of the initial neural network prediction model. Please combine Figure 4 refer to Figure 2 In this embodiment, the method for obtaining an initial neural network prediction model based on the data pairs includes:

[0045] Step S201: using the characteristic parameters that affect the power supply as neurons in the input layer.

[0046] In this embodiment, several characteristic parameters that affect the power supply include: several sets of transformer 208V series, several sets of transformer 308V series, several sets of transformer 480V series, several sets of UPS 208V series, several sets of UPS 380V series, several sets of UPS 480V series, several bus 1600A series, several bus 2000A series, several bus 2500A series and several 2200KW generators.

[0047] Step S202: Obtain the number of nodes in the input layer according to the number of characteristic parameters that affect the power supply amount.

[0048] In this embodiment, the number of nodes in the input layer is obtained according to the number of characteristic parameters that affect the power supply amount, and the number of nodes m in the input layer is 10.

[0049] Step S203: using the power supply as neurons of the output layer, and obtaining the number of nodes of the output layer according to the power supply.

[0050] In this embodiment, the number of nodes in the output layer is obtained according to the power supply amount, and the number n of nodes in the output layer is 1.

[0051] Step S204: Obtain the number of nodes in the hidden layer according to the number of nodes in the input layer and the number of nodes in the output layer.

[0052] The number of nodes in the hidden layer is obtained according to the number of nodes in the input layer and the number of nodes in the output layer. Among them, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

[0053] In this embodiment, the number of nodes m of the input layer is 10, and the number of nodes n of the output layer is 1. Substituting m=10 and n=1 into the formula for calculating the number of nodes h of the hidden layer, the number of nodes h of the hidden layer is obtained to be 6.07. After rounding, the number of nodes h of the hidden layer is 6.

[0054] Step S205: Acquire the initial neural network prediction model according to the neurons of the input layer, the number of nodes of the input layer, the neurons of the output layer, the number of nodes of the output layer, and the number of nodes of the hidden layer.

[0055] The initial neural network prediction model is as follows: Figure 4 The neurons of the input layer include: several sets of transformer 208V series, several sets of transformer 308V series, several sets of transformer 480V series, several sets of UPS 208V series, several sets of UPS 380V series, several sets of UPS 480V series, several bus 1600A series, several bus 2000A series, several bus 2500A series and several 2200KW generators; the neurons of the output layer include: power supply; the number of nodes in the hidden layer is 6.

[0056] In this embodiment, the method for predicting power supply demand also includes: obtaining a connection weight w based on the number of nodes m in the input layer, the number of nodes n in the output layer, and the number of nodes h in the hidden layer; obtaining a hidden layer excitation function, an output layer excitation function, a network training function, and a network performance function.

[0057] The connection weight w is obtained according to the number of nodes m in the input layer, the number of nodes n in the output layer and the number of nodes h in the hidden layer, and the connection weight w=m×n+n×h+n+h, wherein w is the connection weight, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

[0058] In this embodiment, the number of nodes m in the input layer is 10, the number of nodes n in the output layer is 1, and the number of nodes h in the hidden layer is 6. Substituting into the formula for calculating the connection weight w, the connection weight w=10×6+6×1+6+1=73 is obtained.

[0059] In this embodiment, the hidden layer activation function includes the S function (Sigmoid); the output layer activation function includes a linear function; the network training function includes a momentum BP algorithm (traindx) with a variable learning rate; and the network performance function includes a mean square error function (Mean Square Error, mse for short).

[0060] Please continue to refer to Figure 1 , execute step S30: train and test the initial neural network prediction model to obtain a neural network prediction model, and the error of the neural network prediction model is within a preset range.

[0061] Please refer to Figure 3 In this embodiment, the initial neural network prediction model is trained and tested, including:

[0062] Step S301: providing a sample set including a number of data pairs.

[0063] The data pairs include: a plurality of characteristic parameters affecting the power supply amount and the corresponding power supply amount.

[0064] In this embodiment, the data pairs include: several sets of transformer 208V series, several sets of transformer 308V series, several sets of transformer 480V series, several sets of UPS 208V series, several sets of UPS 380V series, several sets of UPS 480V series, several bus 1600A series, several bus 2000A series, several bus 2500A series and several 2200KW generators. As the number of transformers, the number of uninterruptible power supplies, the number of busbars and the number of generators change, the corresponding power supply also changes.

[0065] Please refer to Figure 5 , Figure 5 It is a partial data diagram of several characteristic parameters, where History1, History2, etc. are the values ​​of different groups of characteristic parameters.

[0066] Step S302: Process and group the sample set to obtain training samples, verification samples, and test samples.

[0067] The training samples are used to build the model; the validation samples are used to determine the network structure and adjust the hyperparameters of the model, that is, to quickly adjust the parameters to obtain the optimal model; and the test samples are used to evaluate the performance of the model.

[0068] In this embodiment, the data pairs in the training sample are 50% of the data pairs in the sample set; the data pairs in the validation sample are 25% of the data pairs in the sample set; and the data pairs in the test sample are 25% of the data pairs in the sample set.

[0069] In this embodiment, it is preset that there are 1000 groups of data pairs in the sample set, then there are 500 groups of data pairs in the training sample, there are 250 groups of data pairs in the validation sample, and there are 250 groups of data pairs in the test sample.

[0070] In this embodiment, the tool used to process and group the sample set is Matlab.

[0071] Step S303: Preset the number of iterations, expected error and learning rate, and train the initial neural network prediction model according to the training samples.

[0072] The number of iterations, expected error and learning rate can be adjusted according to actual needs. In this embodiment, the preset number of iterations is 5000 and the expected error is 10 -2 , and the learning rate is 0.01.

[0073] The initial neural network prediction model is trained according to the training samples, including: using 10 characteristic parameters such as several sets of transformer 208V series, several sets of transformer 308V series, several sets of transformer 480V series, several sets of UPS208V series, several sets of UPS380V series, several sets of UPS480V series, several bus 1600A series, several bus 2000A series, several bus 2500A series and several 2200KW generators as the input layer, the input layer nodes are 10, using the power supply as the output layer, the output layer node is 1, the hidden layer nodes are 6, and the initial neural network prediction model is trained with the above data.

[0074] Step S304: Optimizing the initial neural network prediction model according to the verification sample.

[0075] The initial neural network prediction model is optimized according to the verification sample, including: adjusting the hyperparameters of the initial neural network prediction model with the data in the verification sample to obtain an optimal model.

[0076] Step S305: predicting the initial neural network prediction model according to the test sample until the initial neural network prediction model converges, and obtaining a neural network prediction model, wherein the expected error of the neural network prediction model is within a preset range.

[0077] In this embodiment, the expected error of the neural network prediction model is within a preset range, and the expected error of the neural network prediction model is less than 10 -2 .

[0078] Please refer to Figure 6 , Figure 6 This is a graph of neural network performance and error. The horizontal axis is epoch. When a complete data set passes through the neural network once and returns once, this process is called an epoch. Figure 6 The vertical axis is the mean square error (MSE), the curve Train is the image of the mean square error of the training sample changing with the number of training times, the curve Test is the image of the mean square error of the test sample changing with the number of training times, and the curve Validation is the image of the mean square error of the validation sample changing with the number of training times.

[0079] exist Figure 6 In the figure, the dotted line Best indicates that the neural network prediction model converges after 71 training times. At this time, the training effect is the best, and the network error is 1.89x10 -6 , which is less than the expected error of 10 -2 , indicating that the performance of the neural network prediction model is good.

[0080] Please continue to refer to Figure 1 , executing step S40: predicting the demand for power supply according to the neural network prediction model.

[0081] The demand for power supply is predicted according to the neural network prediction model, including: based on the neural network prediction model, according to the demand for power supply, obtaining the number of several sets of transformers with different voltages, the number of several sets of uninterruptible power supplies with different voltages, the number of several busbars with different rated currents, and the number of several generators.

[0082] In this embodiment, according to the demand for the power supply, the number of several sets of transformers with different voltages, the number of several sets of uninterruptible power supplies with different voltages, the number of several buses with different rated currents, and the number of several generators are obtained, including: several sets of transformers of 208V series, several sets of transformers of 308V series, several sets of transformers of 480V series, several sets of UPS208V series, several sets of UPS380V series, several sets of UPS480V series, several busbars of 1600A series, several busbars of 2000A series, several busbars of 2500A series, and several 2200KW generators.

[0083] The method for predicting power supply demand obtains a neural network prediction model based on a number of data pairs consisting of a number of characteristic parameters affecting the power supply and the corresponding power supply. The neural network prediction model can accurately predict the power supply, improve economic benefits and ensure the safe operation of the power system.

[0084] Specifically, from the perspective of economic benefits, power forecasting can effectively avoid excessive power consumption and determine the optimal operation plan for generators, transformers, UPS, busbars, etc. From the perspective of system operation safety, effective power forecasting can formulate a complete maintenance plan and reasonably arrange the number of generators, transformers, UPS, and busbars.

[0085] Accordingly, an embodiment of the present invention further provides a storage medium on which computer instructions are stored, and the computer instructions are executed when they are executed. Figures 1 to 3 The steps of the method.

[0086] Accordingly, an embodiment of the present invention further provides a storage terminal, including a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor, and the processor executes the computer instructions when executing the computer instructions. Figures 1 to 3 The steps of the method.

[0087] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A method for predicting power supply demand, It is characterized in that include: Providing a plurality of data pairs, wherein the plurality of data pairs include a plurality of characteristic parameters affecting the power supply amount and the corresponding power supply amount; Obtaining an initial neural network prediction model based on the plurality of data pairs; Training and testing the initial neural network prediction model to obtain a neural network prediction model, wherein the error of the neural network prediction model is within a preset range; The demand for power supply is predicted based on the neural network prediction model.

2. The method for predicting power supply demand according to claim 1, It is characterized in that The initial neural network prediction model includes an input layer, a hidden layer and an output layer. The method for obtaining the initial neural network prediction model based on several of the data pairs includes: using the several characteristic parameters that affect the power supply as neurons of the input layer; obtaining the number of nodes of the input layer according to the number of characteristic parameters that affect the power supply; using the power supply as neurons of the output layer, and obtaining the number of nodes of the output layer according to the power supply; obtaining the number of nodes of the hidden layer according to the number of nodes of the input layer and the number of nodes of the output layer; obtaining the initial neural network prediction model according to the neurons of the input layer, the number of nodes of the input layer, the neurons of the output layer, the number of nodes of the output layer, and the number of nodes of the hidden layer.

3. The method for predicting power supply demand according to claim 2, It is characterized in that Several characteristic parameters that affect the power supply include: several sets of transformers with different voltages, several sets of uninterruptible power supplies with different voltages, several busbars with different rated currents, and several generators.

4. The method for predicting power supply demand according to claim 3, It is characterized in that Several sets of transformers of different voltages include: several sets of transformers of 208V series, several sets of transformers of 308V series and several sets of transformers of 480V series; several sets of uninterruptible power supplies of different voltages include: several sets of UPS208V series, several sets of UPS380V series and several sets of UPS480V series; several busbars of different rated currents include: several busbars of 1600A series, several busbars of 2000A series and several busbars of 2500A series; several generators include: several 2200KW generators.

5. The method for predicting power supply demand according to claim 4, It is characterized in that The number of nodes of the input layer is obtained according to the number of characteristic parameters affecting the power supply amount, and the number of nodes of the input layer is 10; the number of nodes of the output layer is obtained according to the power supply amount, and the number of nodes of the output layer is 1.

6. The method for predicting power supply demand according to claim 2, It is characterized in that The number of nodes in the hidden layer is obtained according to the number of nodes in the input layer and the number of nodes in the output layer. Among them, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

7. The method for predicting power supply demand according to claim 2, It is characterized in that Before training and testing the initial neural network prediction model, it also includes: obtaining connection weights according to the number of nodes in the input layer, the number of nodes in the output layer, and the number of nodes in the hidden layer; obtaining the hidden layer activation function, the output layer activation function, the network training function, and the network performance function.

8. The method for predicting power supply demand according to claim 7, It is characterized in that The connection weight is obtained according to the number of nodes in the input layer, the number of nodes in the output layer and the number of nodes in the hidden layer, and the connection weight w=m×n+n×h+n+h, wherein w is the connection weight, h is the number of nodes in the hidden layer, m is the number of nodes in the input layer, and n is the number of nodes in the output layer.

9. The method for predicting power supply demand according to claim 7, It is characterized in that The hidden layer excitation function includes an S function; the output layer excitation function includes a linear function; the network training function includes a momentum BP algorithm with a variable learning rate; and the network performance function includes a mean square error function.

10. The method for predicting power supply demand according to claim 7, It is characterized in that The initial neural network prediction model is trained and tested, including: providing a sample set including a number of data pairs; processing and grouping the sample set to obtain training samples, verification samples and test samples; presetting the number of iterations, expected errors and learning rates, and training the initial neural network prediction model according to the training samples; optimizing the initial neural network prediction model according to the verification samples; predicting the initial neural network prediction model according to the test samples until the initial neural network prediction model converges, and obtaining a neural network prediction model, wherein the expected error of the neural network prediction model is within a preset range.

11. The method for predicting power supply demand according to claim 3, It is characterized in that The demand for power supply is predicted according to the neural network prediction model, including: based on the neural network prediction model, according to the demand for power supply, obtaining the number of several sets of transformers with different voltages, the number of several sets of uninterruptible power supplies with different voltages, the number of several busbars with different rated currents, and the number of several generators.

12. A storage medium having computer instructions stored thereon, It is characterized in that When the computer instructions are executed, the steps of the method according to any one of claims 1 to 11 are executed.

13. A storage terminal comprising a memory and a processor, wherein the memory stores computer instructions that can be executed on the processor. It is characterized in that When the processor runs the computer instructions, the steps of the method according to any one of claims 1 to 11 are performed.

Images