A method for predicting copper-iron expansion difference under deep peak regulation of steam turbine generator

Abstract

The invention relates to a method for predicting the copper-iron expansion difference of a steam turbine generator under deep peak-shaving conditions. The technical scheme comprises the following steps: collecting historical data of deep peak-shaving of the generator; optimizing data; performing neural network calculation; and predicting the copper-iron expansion difference of a stator. Compared with manual calculation, the neural network is used to predict the copper-iron expansion difference of a stator under deep peak-shaving conditions, which has high accuracy and fast calculation, and greatly saves time cost. The copper-iron expansion difference data of a generator stator under different deep peak-shaving conditions can be calculated, independent of whether the peak-shaving conditions actually occur. The obtained copper-iron expansion difference data of a generator stator can provide scientific and effective data support for subsequent evaluation of the influence of the copper-iron expansion difference of a generator on the deep peak-shaving capability of the generator.

Description

Technical Field

[0001] The invention relates to a method for predicting copper-iron expansion difference under deep peak regulation conditions of a steam turbine generator. Background Art

[0002] The large-scale access of renewable energy to the power grid has put forward higher requirements for the peak-shaving capacity of the power grid. The participation of thermal power units in deep peak-shaving is the main choice of the current power grid. When the steam turbine generator unit participates in deep peak-shaving, the rapid change of load will cause the copper-iron expansion difference between the stator core and the wire rod. This copper-iron expansion difference will aggravate the loosening of the stator winding and cause failure. The copper-iron expansion difference of the generator body is one of the factors that limit the deep peak-shaving capacity of the generator. At present, the calculation of the copper-iron expansion difference is limited to the temperature data of the generator stator core and stator wire rod before and after deep peak-shaving in the DCS system after the unit participates in deep peak-shaving to calculate the copper-iron expansion difference, which belongs to the category of post-evaluation. When the load of the generator changes too quickly, the copper-iron expansion difference between the stator core and the wire rod may have buried hidden dangers for the safe and stable operation of the generator. If the copper-iron expansion difference under the deep peak-shaving condition of the generator can be predicted, the deep peak-shaving capacity of the generator can be evaluated and restricted according to the predicted data, the operation reliability of the generator will be greatly improved. Summary of the invention

[0003] In view of the above situation, in order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a method for predicting the copper-iron expansion difference of a steam turbine generator under deep peak-shaving conditions, which can predict the copper-iron expansion difference value of the generator under deep peak-shaving conditions, and provide scientific and effective data support for the subsequent evaluation of the impact of the copper-iron expansion difference of the generator on the deep peak-shaving capability of the generator.

[0004] The technical solution provided by the present invention is:

[0005] A method for predicting copper-iron expansion difference under deep peak regulation conditions of a steam turbine generator comprises the following steps:

[0006] Step 1: Collect historical data on deep peak load regulation of generators

[0007] Query the deep peak regulation process of generators in recent years (≥2 years) from the DCS system, collect process data related to deep peak regulation, and establish an original database;

[0008] The process data related to the deep peak regulation include the start time and the end time of the deep peak regulation process, the load of the unit at the beginning of the peak regulation process and the load at the end of the peak regulation process, respectively recorded as the initial load and the end load, the temperature of the stator core at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator core initial temperature and the stator core end temperature, the temperature of the stator wire rod at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator wire rod initial temperature and the stator wire rod end temperature;

[0009] Step 2: Data Optimization

[0010] The process data in the original database is deleted and calculated. The specific method is as follows:

[0011] A. For a certain depth peak regulation process:

[0012] Selection of generator core temperature: The generator core temperature measurement points are arranged along the generator circumference in the middle of the generator axis. After removing the bad points, the temperatures collected by the remaining core measurement points are averaged, and the average temperature is used as the core temperature corresponding to the current power (in degrees Celsius); the generator stator core initial temperature is the core temperature corresponding to the maximum power during the generator power reduction process; the generator stator core end temperature is the core temperature corresponding to the minimum power during the generator power reduction process;

[0013] Generator stator wire bar temperature selection: The generator has multiple wire bars, each of which is equipped with a temperature measuring point. After removing the bad points, the temperatures collected by the remaining wire bar temperature measuring points are averaged, and the average temperature is used as the wire bar temperature corresponding to the current power (in degrees Celsius); the generator stator wire bar initial temperature is the wire bar temperature corresponding to the maximum power during the generator power reduction process; the generator stator wire bar end temperature is the wire bar temperature corresponding to the minimum power during the generator power reduction process;

[0014] Generator core temperature difference ΔT Fe Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator core;

[0015] Generator stator wire temperature difference ΔT Cu Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator stator bar;

[0016] Calculation of the copper-iron expansion difference of the generator stator:

[0017] Stator core expansion ΔL Fe Calculate according to formula (1):

[0018] ΔL Fe =α Fe ×L×ΔT Fe (Formula 1)

[0019] In the formula, ΔL Fe is the temperature change ΔT Fe The change of the stator core length; L is the initial length of the stator core; α Fe is the linear thermal expansion coefficient of iron element, which is 12.2×1E-6 / ℃;

[0020] Generator stator bar expansion ΔL Cu Calculate according to formula (2):

[0021] ΔL Cu =α Cu ×L×ΔT Cu (Formula 2)

[0022] In the formula, ΔL Cu is the temperature change ΔT Cu The change of the length of the lower stator bar; L is the initial length of the stator bar; α Cu is the linear thermal expansion coefficient of copper element, which is 17.5×1E-6 / ℃;

[0023] The expansion difference ΔL between copper and iron in the generator stator is |ΔL Fe -ΔL Cu |;

[0024] Calculation of the duration of generator peak load regulation: The interval between the start and end time of the peak load regulation process is the duration of the peak load regulation. If it is less than 0.5h, it will be calculated as 0.5h.

[0025] Calculation of initial load: Normalize the initial load of the generator, that is, initial load / Pgn, where Pgn is the rated active power of the generator;

[0026] Calculation of end load: Normalize the generator end load, i.e. end load / Pgn, where Pgn is the rated active power of the generator;

[0027] B. Perform the above data screening and calculation for all the deep peak-shaving processes found. Finally, store the generator stator copper-iron expansion difference (in mm), peak-shaving duration (in h), initial load, and end load of each deep peak-shaving process in an Excel table as a preferred database;

[0028] Step 3: Perform neural network calculations

[0029] Matlab software is used to calculate the neural network and the optimal database is imported; a three-layer BP neural network is used, and its input layer is the duration of deep peak regulation, initial load, and end load; the number of hidden layer neuron contacts is determined by trial and error, and the output layer is the copper-iron expansion difference of the generator stator. Finally, the predicted stator copper-iron expansion difference is compared with the actual value and the error is calculated. The threshold and the number of hidden layers are adjusted until the error is minimized, and the neural network is finally determined.

[0030] Step 4: Prediction of stator copper-iron expansion difference

[0031] The relevant deep peak-shaving operating condition data for predicting the copper-iron expansion difference of the generator stator: the duration of deep peak-shaving, the initial load and the end load are filled into the neural network input layer, and the copper-iron expansion difference of the generator stator under this condition can be predicted.

[0032] Compared with the prior art, the method of the present invention has the following beneficial technical effects:

[0033] (1) Compared with manual calculation, the use of neural network to predict the stator copper-iron expansion difference under deep peak-shaving conditions of the generator has high accuracy and fast calculation, which greatly saves time cost.

[0034] (2) The copper-iron expansion difference data of the generator stator under different depth peak-shaving conditions can be calculated, regardless of whether the peak-shaving condition actually occurs.

[0035] (3) The obtained data on the copper-iron expansion difference of the generator stator can provide scientific and effective data support for the subsequent evaluation of the impact of the copper-iron expansion difference of the generator on the deep peak-shaving capability of the generator. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a schematic diagram of a preferred database for an application example of the present invention. DETAILED DESCRIPTION

[0037] The specific implementation modes of the present invention are further described in detail below in conjunction with the embodiments.

[0038] The present invention provides a method for predicting the copper-iron expansion difference under the deep peak regulation condition of a steam turbine generator, comprising the following steps:

[0039] Step 1: Collect historical data on deep peak load regulation of generators

[0040] Query the deep peak regulation process of generators in recent years (≥2 years) from the DCS system, collect process data related to deep peak regulation, and establish an original database;

[0041] The process data related to the deep peak regulation include the start time and the end time of the deep peak regulation process, the load of the unit at the beginning of the peak regulation process and the load at the end of the peak regulation process, respectively recorded as the initial load and the end load, the temperature of the stator core at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator core initial temperature and the stator core end temperature, the temperature of the stator wire rod at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator wire rod initial temperature and the stator wire rod end temperature;

[0042] The deep peak-shaving process is defined as: when the active power value of the generator decreases continuously from the power P greater than 0.6Pgn to less than 0.4Pgn, and then the active power of the generator continues to rise to greater than 0.4Pgn, and the generator power reduction process time is less than 8h, then the power reduction process is the deep peak-shaving process of the generator; wherein Pgn is the rated active power of the generator.

[0043] Step 2: Data Optimization

[0044] The process data in the original database is deleted and calculated. The specific method is as follows:

[0045] A. For a certain depth peak regulation process:

[0046] Selection of generator core temperature: The generator core temperature measurement points are arranged along the generator circumference in the middle of the generator axis. After removing the bad points, the temperatures collected by the remaining core measurement points are averaged, and the average temperature is used as the core temperature corresponding to the current power (in degrees Celsius); the generator stator core initial temperature is the core temperature corresponding to the maximum power during the generator power reduction process; the generator stator core end temperature is the core temperature corresponding to the minimum power during the generator power reduction process;

[0047] Generator stator wire bar temperature selection: The generator has multiple wire bars, each of which is equipped with a temperature measuring point. After removing the bad points, the temperatures collected by the remaining wire bar temperature measuring points are averaged, and the average temperature is used as the wire bar temperature corresponding to the current power (in degrees Celsius); the generator stator wire bar initial temperature is the wire bar temperature corresponding to the maximum power during the generator power reduction process; the generator stator wire bar end temperature is the wire bar temperature corresponding to the minimum power during the generator power reduction process;

[0048] Generator core temperature difference ΔT Fe Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator core;

[0049] Generator stator wire temperature difference ΔT Cu Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator stator bar;

[0050] Calculation of the copper-iron expansion difference of the generator stator:

[0051] Stator core expansion ΔL Fe Calculate according to formula (1):

[0052] ΔL Fe =α Fe ×L×ΔT Fe (Formula 1)

[0053] In the formula, ΔL Fe is the temperature change ΔT Fe The change of the stator core length; L is the initial length of the stator core, 6.7 (taking 660MW unit as an example); α Fe is the linear thermal expansion coefficient of iron element, which is 12.2×1E-6 / ℃;

[0054] Generator stator bar expansion ΔL Cu Calculate according to formula (2):

[0055] ΔL Cu =αCu ×L×ΔT Cu (Formula 2)

[0056] In the formula, ΔL Cu is the temperature change ΔT Cu The change of the length of the lower stator bar; L is the initial length of the stator bar, which is 10.5 meters (taking the 660MW unit as an example); α Cu is the linear thermal expansion coefficient of copper element, which is 17.5×1E-6 / ℃;

[0057] The expansion difference ΔL between copper and iron in the generator stator is |ΔL Fe -ΔL Cu |;

[0058] Calculation of the duration of generator peak load regulation: The interval between the start and end time of the peak load regulation process is the duration of the peak load regulation. If it is less than 0.5h, it will be calculated as 0.5h.

[0059] Calculation of initial load: Normalize the initial load of the generator, that is, initial load / Pgn, where Pgn is the rated active power of the generator;

[0060] Calculation of end load: Normalize the generator end load, i.e. end load / Pgn, where Pgn is the rated active power of the generator;

[0061] B. Perform the above data screening and calculation for all the deep peak-shaving processes found. Finally, store the generator stator copper-iron expansion difference (in mm), peak-shaving duration (in h), initial load, and end load of each deep peak-shaving process in an Excel table as a preferred database;

[0062] Step 3: Perform neural network calculations

[0063] Matlab software is used to calculate the neural network and the optimal database is imported; a three-layer BP neural network is used, and its input layer is the duration of deep peak regulation, initial load, and end load; the number of hidden layer neuron contacts is determined by trial and error, and the output layer is the copper-iron expansion difference of the generator stator. Finally, the predicted stator copper-iron expansion difference is compared with the actual value and the error is calculated. The threshold and the number of hidden layers are adjusted until the error is minimized, and the neural network is finally determined.

[0064] Step 4: Prediction of stator copper-iron expansion difference

[0065] The relevant deep peak-shaving operating condition data for predicting the copper-iron expansion difference of the generator stator: the duration of deep peak-shaving, the initial load and the end load are filled into the neural network input layer, and the copper-iron expansion difference of the generator stator under this condition can be predicted.

[0066] The present invention has achieved good technical effects through practical application, and the application examples are as follows:

[0067] Step 1: Collect data:

[0068] The deep peak-shaving process in which the generators participated in the past five years was queried from the DCS system of a power plant, and relevant deep peak-shaving process data was collected.

[0069] Step 2: Establish a preferred database

[0070] The data collected in step 1 is established in the original database, and the original database is deleted and calculated according to step 2. The data is stored in Excel format to obtain the optimal database. In this embodiment, 32 sets of peak-shaving data are obtained by the above method, such as Figure 1 As shown, column 1 is the copper-iron expansion difference of the generator stator (in mm), column 2 is the duration of the generator deep peak regulation (in h), column 3 is the initial load, and column 4 is the end load.

[0071] Step 3: BP neural network programming and calculation:

[0072] (1) Open the matlab software and import the data in the optimal database into the ".mat" format table in the software as the data source for BP neural network training and testing;

[0073] (2) Create a new ".m" format and perform neural network programming. This experiment establishes a three-layer BP neural network: input layer, hidden layer, and output layer. The input layer is the peak load duration, initial load, and end load; the number of hidden layer neuron contacts is 12, and the output layer is the expansion difference of copper and iron in the generator stator. The obtained neural network is:

[0074] net=newff(minmax(P),[3,12,1],{'tansig','tansig','purelin'},'trainlm')

[0075] The first 22 data sets were used for training, and the 5 data sets from 23 to 27 were used for testing. The threshold of the neural network was gradually adjusted by observing the error to obtain the optimal network.

[0076] Step 4: Prediction of the copper-iron expansion difference of the generator stator under different deep adjustment conditions:

[0077] The predictions for groups 28-32 are made, and the comparison between the prediction results and the actual copper-iron expansion difference data of the stator is shown in Table 1.

[0078] Table 1 Comparison of predicted and actual values ​​of data sets 28-32

[0079] Group i 28 29 30 31 32 Predicting the Differential Expansion of Copper and Iron 0.7356 0.5632 0.2722 0.4416 0.9678 Actual copper and iron expansion difference 0.7678 0.5784 0.2631 0.4321 0.9446 error(%) 4.19 2.63 3.46 2.20 2.46

[0080] It can be seen from Table 1 that the prediction error is within 5%, the model accuracy is good, and it meets the needs of actual field application.

Claims

1. A method for predicting the copper-iron expansion difference under deep peak regulation conditions of a steam turbine generator, characterized in that: The following steps are involved: Step 1: Collect historical data on deep peak load regulation of generators Query the deep peak regulation process of generators in recent years from the DCS system, collect process data related to deep peak regulation, and establish an original database; The process data related to the deep peak regulation include the start time and the end time of the deep peak regulation process, the load of the unit at the beginning of the peak regulation process and the load at the end of the peak regulation process, respectively recorded as the initial load and the end load, the temperature of the stator core at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator core initial temperature and the stator core end temperature, the temperature of the stator wire rod at the beginning of the peak regulation process and the temperature at the end of the peak regulation process, respectively recorded as the stator wire rod initial temperature and the stator wire rod end temperature; Step 2: Data Optimization The process data in the original database is deleted and calculated. The specific method is as follows: A. For a certain depth peak regulation process: Selection of generator core temperature: The generator core temperature measurement points are arranged along the generator circumference in the middle of the generator axis. After removing the bad points, the temperatures collected by the remaining core measurement points are averaged, and the average temperature is used as the core temperature corresponding to the current power; the generator stator core initial temperature is the core temperature corresponding to the maximum power during the generator power reduction process; the generator stator core end temperature is the core temperature corresponding to the minimum power during the generator power reduction process; Selection of generator stator wire bar temperature: The generator has multiple wire bars, each of which is equipped with a temperature measuring point. After removing the bad points, the temperatures collected by the remaining wire bar temperature measuring points are averaged, and the average temperature is used as the wire bar temperature corresponding to the current power; the generator stator wire bar initial temperature is the wire bar temperature corresponding to the maximum power during the generator power reduction process; the generator stator wire bar end temperature is the wire bar temperature corresponding to the minimum power during the generator power reduction process; Generator core temperature difference ΔT Fe Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator core; Generator stator wire temperature difference ΔT Cu Calculation: The absolute value of the temperature difference between the initial temperature and the final temperature of the generator stator bar; Calculation of the copper-iron expansion difference of the generator stator: Stator core expansion ΔL Fe Calculate according to formula (1): ΔL Fe = α Fe × L × ΔT Fe (Equation 1) Where, ΔL Fe is the temperature change ΔT Fe The change of the stator core length; L is the initial length of the stator core; α Fe is the linear thermal expansion coefficient of iron element, which is 12.2×1E-6 / ℃; Generator stator bar expansion ΔL Cu Calculate according to formula (2): ΔL Cu = α Cu × L × ΔT Cu (Equation 2) Where, ΔL Cu is the temperature change ΔT Cu The change of the length of the lower stator bar; L is the initial length of the stator bar; α Cu is the linear thermal expansion coefficient of copper element, which is 17.5×1E-6 / ℃; The expansion difference ΔL between copper and iron in the generator stator is |ΔL Fe -ΔL Cu |; Calculation of the duration of generator peak load regulation: The interval between the start and end time of the peak load regulation process is the duration of the peak load regulation. If it is less than 0.5h, it will be calculated as 0.5h. Calculation of initial load: Normalize the initial load of the generator, that is, initial load / Pgn, where Pgn is the rated active power of the generator; Calculation of end load: Normalize the generator end load, i.e. end load / Pgn, where Pgn is the rated active power of the generator; B. Perform the above data screening and calculation for all the deep peak-shaving processes found. Finally, store the generator stator copper-iron expansion difference, peak-shaving duration, initial load, and end load of each deep peak-shaving process in an Excel spreadsheet as a preferred database. Step 3: Perform neural network calculations Matlab software is used to calculate the neural network and import the optimal database; a three-layer BP neural network is used, and its input layer is the duration of deep peak regulation, initial load, and end load; the number of hidden layer neuron contacts is determined by trial and error, and the output layer is the copper-iron expansion difference of the generator stator. Finally, the predicted copper-iron expansion difference of the stator is compared with the actual value and the error is calculated. The threshold and the number of hidden layers are adjusted until the error is minimized, and the neural network is finally determined; Step 4: Prediction of stator copper-iron expansion difference The relevant deep peak-shaving operating condition data for predicting the copper-iron expansion difference of the generator stator: the duration of deep peak-shaving, the initial load and the end load are filled into the neural network input layer, and the copper-iron expansion difference of the generator stator under this condition can be predicted.

2. The method for predicting the copper-iron expansion difference under deep peak load conditions of a steam turbine generator according to claim 1, characterized in that: The deep peak-shaving process is defined as: when the active power value of the generator decreases continuously from the power P greater than 0.6Pgn to less than 0.4Pgn, and then the active power of the generator continues to rise to greater than 0.4Pgn, and the generator power reduction process time is less than 8h, then the power reduction process is the deep peak-shaving process of the generator; wherein Pgn is the rated active power of the generator.

Images