Interpolation method based on wind lidar data

Abstract

The invention discloses an interpolation method based on wind measurement laser radar data, and the method comprises the following steps: S1, determining a to-be-measured radar and a wind measurement device, setting the wind measurement radar as a wind measurement device adjacent to the to-be-measured radar, and setting the wind measurement radar as a wind measurement tower or a normally operating radar closest to the to-be-measured radar; s2, wind speed data of N height layers of the day before interpolation of the radar to be measured and the wind measurement equipment are acquired at the same time, wherein the wind speed data comprise the X-direction wind speed, the Y-direction wind speed and the Z-direction wind speed; dividing the wind speed data of the N height layers of the radar to be measured and the wind measurement equipment; s3, judging whether the wind speed data is valid data or not through the wind speed flag bit, screening out height layer data of the to-be-measured radar and the wind measurement equipment which are valid wind speed data, and forming a 3 * N-dimensional feature data set; compared with the prior art, the defect of data missing of the wind measurement laser radar under the abnormal working condition can be made up; the method combines the distribution characteristics of the wind speed in time and space in a complex terrain environment, and guarantees the accuracy and reliability of the wind measurement laser radar interpolation data.

Description

Technical Field

[0001] The present invention relates to the field of laser radar measurement, and in particular to an interpolation method based on wind laser radar data. Background Art

[0002] Wind farm wind measurement data has a direct and important impact on the assessment of wind farm capacity. According to the "Wind Energy Resource Assessment Method for Wind Farms" (GB / T 18710-2002), wind measurement data needs to have a complete annual measurement time. However, in actual situations, due to wind laser radar failures, abnormal weather and other reasons, there are often some invalid data in the wind laser radar data. In order to objectively reflect the wind resource situation of the wind farm, we need to clean and complete the wind laser radar data to achieve a complete 1-year data. In addition, wind measurement work occupies a lot of time in the early stage of the wind farm project. Therefore, when the actual measurement time is short, providing a reasonable and fast data completion method will help to obtain a complete 1-year data faster, and then conduct a preliminary assessment of the wind resources of the wind farm project and improve the project development efficiency. At present, the existing data interpolation methods include linear interpolation, interpolation, statistical interpolation methods, spatiotemporal interpolation, etc. It is difficult to take into account the complex terrain or complex spatial distribution of wind farm scenes, and it is difficult to ensure the accuracy and reliability of the interpolated data. Summary of the invention

[0003] The purpose of the present invention is to provide a wind measurement data interpolation method based on wind measurement lidar data, which requires a lot of measurement time and contains some invalid data. In the process of cleaning and completing wind measurement lidar data, the existing methods are difficult to take into account complex terrain or complex spatially distributed wind field scenes, and it is difficult to ensure the accuracy and reliability of the interpolated data. To address this deficiency, an interpolation method based on wind measurement lidar data is proposed.

[0004] In order to achieve the above object, the present invention adopts the following technical solution:

[0005] An interpolation method based on wind laser radar data comprises the following steps:

[0006] S1. Confirm the radar to be tested and the wind measuring device. The wind measuring radar is set as a wind measuring device adjacent to the radar to be tested, and the wind measuring radar is set as a wind measuring tower or a normally operating radar closest to the radar to be tested;

[0007] S2, simultaneously obtain the wind speed data of N height layers of the radar to be tested and the wind measuring device to interpolate the previous day, wherein the wind speed data includes wind speed in the X direction, wind speed in the Y direction, and wind speed in the Z direction; and divide the wind speed data of the N height layers of the radar to be tested and the wind measuring device respectively;

[0008] S3, judging whether the wind speed data is valid data through the wind speed flag, and screening out the height layer data for which the radar and wind measuring equipment to be tested are valid wind speed data, and forming a 3*N dimension feature data set;

[0009] S4, calculating the measured value of horizontal wind speed and the measured value of horizontal wind direction at any altitude layer of the radar to be measured at any time;

[0010] S5, based on the heights of any two adjacent height layers of the radar to be measured, combined with the measured horizontal wind speed values ​​of the two adjacent height layers, calculate the shear index of the two adjacent height layers at the current moment;

[0011] S6. Along the time dimension, where the time dimension has M moments and N-1 shear indices at the same moment, calculate the average value α of the shear indices of the same two adjacent height layers at M moments. m , m=2,3,4,…,N;

[0012] S7, normalizing the 3*N dimensional feature data sets of the radar to be tested and the wind measuring device according to the feature dimensions;

[0013] S8, dividing the normalized 3*N-dimensional feature data sets of the radar to be tested and the wind measuring device into a training sample set and a test sample set in a ratio of 8:2 along the time dimension direction;

[0014] S9, interpolating the wind speed data of N height layers of the previous day by the radar to be tested and the wind measuring equipment as the input data set and the expected output data set, respectively, and constructing a prediction model based on the neural network model of the multi-layer perceptron;

[0015] S10, optimizing the prediction model by combining the training sample set and the test sample set, and performing wind speed prediction by using the optimized prediction model to form a wind speed prediction set;

[0016] S11. Determine the wind speed data of all altitude layers and output the wind speed.

[0017] As a further preferred embodiment of the present invention, the method of judging whether the wind speed data is valid data by the wind speed flag bit includes:

[0018] When the wind speed flag is 1, the wind speed is valid, and the valid wind speed data is non-Nan;

[0019] When the wind speed flag is 0, the wind speed is invalid, and the invalid wind speed data is Nan.

[0020] As a further preferred embodiment of the present invention, the prediction model is optimized by combining the training sample set and the test sample set, and the wind speed prediction is performed by the optimized prediction model to form a wind speed prediction set, including:

[0021] S101, training the prediction model based on the training sample set, setting the maximum number of iterations, setting the loss function as the mean square error, and optimizing the prediction model parameters using an adaptive moment estimation optimizer until the mean square error between the wind speed prediction value and the radar measured wind speed value is minimized, thereby optimizing the prediction model;

[0022] S102, evaluating the prediction model based on the test sample set, applying the prediction model to the test sample set to obtain a wind speed prediction value, calculating the mean square error between the wind speed prediction value and the radar measured wind speed value, and further optimizing the prediction model;

[0023] S103, based on the optimized prediction model, combined with the wind speed data of all altitude layers of the day by the wind measuring laser radar, obtain the wind speed prediction data of all altitude layers of the day by the radar to be tested.

[0024] As a further preferred embodiment of the present invention, the wind speed data of all altitude layers are judged and the wind speed is output, including:

[0025] S111, obtain the measured values ​​of horizontal wind speed at M times, wind speed forecast data and the average value α of the shear index of the same two adjacent height layers at N height layers of the radar to be measured on the interpolated day m ;

[0026] S112, judging whether the measured value of the horizontal wind speed of the first altitude layer at the current moment is valid;

[0027] S113. If the measured value of the horizontal wind speed at the first altitude layer is valid, output the measured value of the horizontal wind speed at the first altitude layer;

[0028] S114, if the measured value of the horizontal wind speed of the first altitude layer is invalid, determine whether the wind speed prediction accuracy of the first altitude layer at the current moment is greater than a threshold; the wind speed prediction accuracy is the standard deviation between the wind speed prediction data and the measured value of the horizontal wind speed;

[0029] S115, if the wind speed prediction accuracy of the first altitude layer at the current moment is less than or equal to the threshold, determine whether the horizontal wind speed at the moment before the current moment is valid; if valid, jump to S116; if invalid, jump to S115;

[0030] S116, outputting the effective measured value of horizontal wind speed at the last moment;

[0031] S117, if the wind speed prediction accuracy of the first altitude layer at the current moment is greater than the threshold, output the wind speed prediction data of the first altitude layer;

[0032] S118, judging whether the actual measured value of the horizontal wind speed at the nth altitude layer at the current moment is valid, n=2, 3, 4, ..., N;

[0033] S119. If the measured value of the horizontal wind speed at the n-th altitude layer is valid, output the measured value of the horizontal wind speed at the n-th altitude layer;

[0034] S110, if the measured value of the horizontal wind speed at the nth altitude layer is invalid, determine whether the measured value of the horizontal wind speed at the n-1th altitude layer is valid;

[0035] S1111. If the measured value of the horizontal wind speed at the n-1th altitude layer is valid, the horizontal wind speed at the nth altitude layer is calculated based on the average shear index between the nth altitude layer and the previous altitude layer, and outputted;

[0036] S1112. If the measured value of the horizontal wind speed at the n-1th altitude layer is invalid, determine whether the wind speed prediction accuracy at the nth altitude layer at the current moment is greater than a threshold;

[0037] S1113. If the wind speed prediction accuracy of the nth altitude layer at the current moment is less than or equal to the threshold, the horizontal wind speed of the nth altitude layer is calculated according to the average shear index of the nth altitude layer and the n-1th altitude layer, and outputted;

[0038] S1114. If the wind speed prediction accuracy of the nth altitude layer at the current moment is greater than the threshold, output the wind speed prediction data of the nth altitude layer;

[0039] S1115. Circularly determine the effective wind speeds of N altitude layers at M times and output them.

[0040] As a further preferred embodiment of the present invention, the threshold value is 0.1 m / s-1.0 m / s.

[0041] The interpolation method based on wind laser radar data proposed in the present invention has the following beneficial effects compared with the prior art:

[0042] 1. The interpolation method of the present invention can make up for the deficiency of missing data of wind laser radar under abnormal working conditions;

[0043] 2. The interpolation method involved in the present invention, combined with the distribution characteristics of wind speed in time and space in a complex terrain environment, ensures the accuracy and reliability of the interpolation data of the wind laser radar. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the process of the present invention;

[0045] Figure 2 It is the distribution of the predicted value and the reference value of the horizontal wind speed at a certain altitude layer;

[0046] Figure 3 It is a schematic diagram of the wind speed data screening process for all altitude layers. DETAILED DESCRIPTION

[0047] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] An interpolation method based on wind laser radar data comprises the following steps:

[0049] S1. Confirm the radar to be tested and the wind measuring device. The wind measuring radar is set as a wind measuring device adjacent to the radar to be tested, and the wind measuring radar is set as a wind measuring tower or a normally operating radar closest to the radar to be tested.

[0050] S2. Simultaneously obtain the wind speed data of N height layers of the radar to be tested and the wind measuring device to interpolate the previous day, wherein the wind speed data includes wind speed in the X direction, wind speed in the Y direction, and wind speed in the Z direction; and divide the wind speed data of the N height layers of the radar to be tested and the wind measuring device respectively.

[0051] S3. Determine whether the wind speed data is valid data through the wind speed flag, and select the height layer data for which the radar and wind measuring equipment to be tested are valid wind speed data, and form a 3*N dimensional feature data set.

[0052] When the wind speed flag is 1, the wind speed is valid, and the valid wind speed data is non-Nan;

[0053] When the wind speed flag is 0, the wind speed is invalid, and the invalid wind speed data is Nan.

[0054] S4. Calculate the measured value of the horizontal wind speed and the measured value of the horizontal wind direction at any altitude layer of the radar to be tested at any time.

[0055] S5. Based on the heights of any two adjacent height layers of the radar to be tested and in combination with the measured horizontal wind speed values ​​of the two adjacent height layers, the shear index of the two adjacent height layers at the current moment is calculated.

[0056] S6. Along the time dimension, where the time dimension has M moments and N-1 shear indices at the same moment, calculate the average value α of the shear indices of the same two adjacent height layers at M moments. m , m=2,3,4,…,N.

[0057] S7. Normalize the 3*N dimensional feature data sets of the radar to be tested and the wind measuring device according to the feature dimensions.

[0058] S8. The normalized 3*N-dimensional feature data sets of the radar to be tested and the wind measuring device are divided into a training sample set and a test sample set in a ratio of 8:2 along the time dimension direction.

[0059] S9. The wind speed data of N height layers of the previous day interpolated by the radar and wind measuring equipment to be tested are set as the input data set and the expected output data set respectively, and a prediction model is constructed based on the neural network model of the multi-layer perceptron.

[0060] S10. Optimizing the prediction model by combining the training sample set and the test sample set, and performing wind speed prediction by using the optimized prediction model to form a wind speed prediction set.

[0061] The specific steps include:

[0062] S101. Train the prediction model based on the training sample set, set the maximum number of iterations, set the loss function to mean square error, and use an adaptive moment estimation optimizer to optimize the prediction model parameters until the mean square error between the wind speed prediction value and the radar measured wind speed value is minimized, thereby achieving optimization of the prediction model.

[0063] The parameters of the neural network model based on the multi-layer perceptron include the maximum number of iterations, loss function, and optimizer. These parameters are set to achieve the minimum mean square error between the wind speed prediction value and the expected output value faster and better.

[0064] S102, evaluating the prediction model based on the test sample set, applying the prediction model to the test sample set to obtain a wind speed prediction value, calculating the mean square error between the wind speed prediction value and the radar measured wind speed value, and further optimizing the prediction model.

[0065] The test sample set is used to evaluate the accuracy of the prediction model. When the mean square error between the predicted wind speed value and the actual wind speed value measured by the radar is greater than the threshold, the threshold is selected as 0.1m / s-1.0m / s, the confidence of the prediction model is high, and the predicted wind speed value can be used to interpolate the missing horizontal wind speed; otherwise, the missing horizontal wind speed is interpolated directly based on the shear index and the wind speed value of the previous altitude layer. Figure 2 The distribution of the predicted and reference values ​​of horizontal wind speed at a certain altitude is shown.

[0066] S103, based on the optimized prediction model, combined with the wind speed data of all altitude layers of the day by the wind measuring laser radar, obtain the wind speed prediction data of all altitude layers of the day by the radar to be tested.

[0067] The prediction model outputs the wind speed in the X, Y, and Z directions. Combined with the effective wind speed values ​​in the X, Y, and Z directions of all altitude layers at the current moment of the wind measuring lidar, the prediction model is applied to obtain the predicted values ​​of the wind speed in the X, Y, and Z directions. Combined with the synthesis formula of the horizontal wind speed, the predicted values ​​of the horizontal wind speed at all altitude layers are obtained. Based on the measured value of the horizontal wind speed, the predicted value of the horizontal wind speed, and the wind shear index, the horizontal wind speed values ​​of all altitude layers are output.

[0068] S11. Determine the wind speed data of all altitude layers and output the wind speed.

[0069] The specific steps include:

[0070] S111, obtain the measured values ​​of horizontal wind speed at M times, wind speed forecast data and the average value α of the shear index of the same two adjacent height layers at N height layers of the radar to be measured on the interpolated day m .

[0071] S112. Determine whether the actual measured value of the horizontal wind speed of the first altitude layer at the current moment is valid.

[0072] S113. If the measured value of the horizontal wind speed at the first altitude layer is valid, output the measured value of the horizontal wind speed at the first altitude layer.

[0073] S114, if the measured value of the horizontal wind speed of the first altitude layer is invalid, determine whether the wind speed prediction accuracy of the first altitude layer at the current moment is greater than a threshold; the wind speed prediction accuracy is the standard deviation between the wind speed prediction data and the measured value of the horizontal wind speed;

[0074] S115. If the wind speed prediction accuracy of the first altitude layer at the current moment is less than or equal to the threshold, determine whether the horizontal wind speed at the moment before the current moment is valid; if valid, jump to S116; if invalid, jump to S115.

[0075] S116. Output the actual measured value of horizontal wind speed that was valid at the last moment.

[0076] S117: If the wind speed prediction accuracy of the first altitude layer at the current moment is greater than a threshold, the wind speed prediction data of the first altitude layer is output.

[0077] S118. Determine whether the actual measured value of the horizontal wind speed at the nth altitude layer at the current moment is valid, where n=2, 3, 4, ..., N.

[0078] S119. If the measured value of the horizontal wind speed at the nth altitude layer is valid, the measured value of the horizontal wind speed at the nth altitude layer is output.

[0079] S1110. If the measured value of the horizontal wind speed at the nth altitude layer is invalid, determine whether the measured value of the horizontal wind speed at the n-1th altitude layer is valid.

[0080] S1111. If the measured value of the horizontal wind speed at the n-1th altitude layer is valid, the horizontal wind speed at the nth altitude layer is calculated based on the average shear index between the nth altitude layer and the previous altitude layer, and the result is output.

[0081] S1112. If the measured value of the horizontal wind speed at the n-1th altitude layer is invalid, determine whether the wind speed prediction accuracy at the nth altitude layer at the current moment is greater than a threshold.

[0082] S1113. If the wind speed prediction accuracy of the nth altitude layer at the current moment is less than or equal to the threshold, the horizontal wind speed of the nth altitude layer is calculated based on the average shear index of the nth altitude layer and the n-1th altitude layer, and output.

[0083] S1114. If the wind speed prediction accuracy of the nth altitude layer at the current moment is greater than a threshold, the wind speed prediction data of the nth altitude layer is output.

[0084] S1115. Circularly determine the effective wind speeds of N altitude layers at M times and output them.

[0085] Example 2: Collect wind radar data, with 0:00:00 as the first moment, collect data once every second, and there are a total of N=12 height layers. The wind speed data at the first moment is shown in Table 1

[0086] Table 1 shows the wind speed in the X, Y, and Z directions of each altitude layer collected at the first moment

[0087]

[0088] Calculate the horizontal wind speed at each altitude layer and the shear index of the two adjacent altitude layers at the current moment, see Table 2.

[0089] Table 2 shows the horizontal wind speed at each altitude layer at the first moment and the shear index of the two adjacent altitude layers

[0090]

[0091]

[0092] The wind speed data predicted by the prediction model are shown in Table 3.

[0093] Table 3 shows the wind speed data of all altitude layers at the first moment predicted by the prediction model.

[0094]

[0095]

[0096] At the first moment, the measured wind speed at the first altitude layer is invalid, and the standard deviation between the wind speed prediction data and the measured horizontal wind speed is calculated.

[0097] The standard deviation is less than the threshold of 0.5 m / s. The wind speed data of the previous day from the wind radar is retrieved, and the horizontal wind speeds of the first altitude layer at 23:59:59, 23:59:58, 23:59:57, etc. are selected in turn, as shown in Table 4.

[0098] Table 4 shows the horizontal wind speed at the first altitude layer on the previous day.

[0099] time Level Horizontal wind speed (m / s) Is it effective? 23:59:59 First altitude layer Nan no 23:59:58 First altitude layer 5.40 yes 23:59:57 First altitude layer 5.25 yes ......

[0100] The effective wind speed data closest to the first moment, that is, the horizontal wind speed at 23:59:58, is selected, and the horizontal wind speed at 23:59:58 is output as the wind speed data at the first moment of the first altitude layer.

[0101] If the wind speed data of the second altitude layer is valid, it will be directly output. The wind speed data of all altitude layers will be judged in sequence and output.

[0102] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any form, and any technical solution obtained by equivalent replacement or equivalent transformation falls within the protection scope of the present invention.

Claims

1. An interpolation method based on wind laser radar data, characterized in that: The following steps are involved: S1. Confirm the radar to be tested and the wind measuring device. The wind measuring radar is set as a wind measuring device adjacent to the radar to be tested, and the wind measuring radar is set as a wind measuring tower or a normally operating radar closest to the radar to be tested; S2, simultaneously obtain the wind speed data of N height layers of the radar to be tested and the wind measuring device to interpolate the previous day, wherein the wind speed data includes wind speed in the X direction, wind speed in the Y direction, and wind speed in the Z direction; and divide the wind speed data of the N height layers of the radar to be tested and the wind measuring device respectively; S3, judging whether the wind speed data is valid data through the wind speed flag, and screening out the height layer data for which the radar and wind measuring equipment to be tested are valid wind speed data, and forming a 3*N dimension feature data set; S4, calculating the measured value of horizontal wind speed and the measured value of horizontal wind direction at any altitude layer of the radar to be measured at any time; S5, based on the heights of any two adjacent height layers of the radar to be measured, combined with the measured horizontal wind speed values ​​of the two adjacent height layers, calculate the shear index of the two adjacent height layers at the current moment; S6. Along the time dimension, where the time dimension has M moments and N-1 shear indices at the same moment, calculate the average shear index of the same two adjacent height layers at M moments. , m=2,3,4,…,N; S7, normalizing the 3*N dimensional feature data sets of the radar to be tested and the wind measuring device according to the feature dimensions; S8, dividing the normalized 3*N-dimensional feature data sets of the radar to be tested and the wind measuring device into a training sample set and a test sample set in a ratio of 8:2 along the time dimension direction; S9, interpolating the wind speed data of N height layers of the previous day by the radar to be tested and the wind measuring equipment as the input data set and the expected output data set, respectively, and constructing a prediction model based on the neural network model of the multi-layer perceptron; S10, optimizing the prediction model by combining the training sample set and the test sample set, and performing wind speed prediction by using the optimized prediction model to form a wind speed prediction set; S11. Determine the wind speed data of all altitude layers and output the wind speed.

2. The interpolation method based on wind laser radar data according to claim 1 is characterized in that: The method of judging whether the wind speed data is valid data by the wind speed flag bit includes: When the wind speed flag is 1, the wind speed is valid, and the valid wind speed data is non-Nan; When the wind speed flag is 0, the wind speed is invalid, and the invalid wind speed data is Nan.

3. The interpolation method based on wind laser radar data according to claim 2 is characterized in that: The prediction model is optimized by combining the training sample set and the test sample set, and the wind speed is predicted by the optimized prediction model to form a wind speed prediction set, including: S101, training the prediction model based on the training sample set, setting the maximum number of iterations, setting the loss function as the mean square error, and optimizing the prediction model parameters using an adaptive moment estimation optimizer until the mean square error between the wind speed prediction value and the radar measured wind speed value is minimized, thereby optimizing the prediction model; S102, evaluating the prediction model based on the test sample set, applying the prediction model to the test sample set to obtain a wind speed prediction value, calculating the mean square error between the wind speed prediction value and the radar measured wind speed value, and further optimizing the prediction model; S103, based on the optimized prediction model, combined with the wind speed data of all altitude layers of the day by the wind measuring laser radar, obtain the wind speed prediction data of all altitude layers of the day by the radar to be tested.

4. The interpolation method based on wind laser radar data according to claim 2 is characterized in that: The wind speed data of all altitude layers are judged and the wind speed is output, including: S111, obtain the measured values ​​of horizontal wind speed at M times at N height layers of the radar to be measured on the interpolated day, the wind speed forecast data, and the average value of the shear index of the same two adjacent height layers ; S112, judging whether the measured value of the horizontal wind speed of the first altitude layer at the current moment is valid; S113. If the measured value of the horizontal wind speed at the first altitude layer is valid, output the measured value of the horizontal wind speed at the first altitude layer; S114, if the measured value of the horizontal wind speed of the first altitude layer is invalid, determine whether the wind speed prediction accuracy of the first altitude layer at the current moment is greater than a threshold; the wind speed prediction accuracy is the standard deviation between the wind speed prediction data and the measured value of the horizontal wind speed; S115, if the wind speed prediction accuracy of the first altitude layer at the current moment is less than or equal to the threshold, determine whether the horizontal wind speed at the moment before the current moment is valid; if valid, jump to S116; if invalid, jump to S115; S116, outputting the effective measured value of horizontal wind speed at the last moment; S117, if the wind speed prediction accuracy of the first altitude layer at the current moment is greater than the threshold, output the wind speed prediction data of the first altitude layer; S118, judging whether the measured value of the horizontal wind speed at the nth altitude layer at the current moment is valid, n=2, 3, 4, ..., N; S119. If the measured value of the horizontal wind speed at the n-th altitude layer is valid, output the measured value of the horizontal wind speed at the n-th altitude layer; S110, if the measured value of the horizontal wind speed at the nth altitude layer is invalid, determine whether the measured value of the horizontal wind speed at the n-1th altitude layer is valid; S1111. If the measured value of the horizontal wind speed at the n-1th altitude layer is valid, the horizontal wind speed at the nth altitude layer is calculated based on the average shear index between the nth altitude layer and the previous altitude layer, and outputted; S1112. If the measured value of the horizontal wind speed at the n-1th altitude layer is invalid, determine whether the wind speed prediction accuracy at the nth altitude layer at the current moment is greater than a threshold; S1113. If the wind speed prediction accuracy of the nth altitude layer at the current moment is less than or equal to the threshold, the horizontal wind speed of the nth altitude layer is calculated according to the average shear index of the nth altitude layer and the n-1th altitude layer, and outputted; S1114. If the wind speed prediction accuracy of the nth altitude layer at the current moment is greater than a threshold, the wind speed prediction data of the nth altitude layer is output.

5. The interpolation method based on wind laser radar data according to claim 4 is characterized in that: The threshold value is 0.1m / s-1.0m / s.

Images