CN106997407B - Wind resource scene reduction method based on trend fitting - Google Patents
Wind resource scene reduction method based on trend fitting Download PDFInfo
- Publication number
- CN106997407B CN106997407B CN201611248741.7A CN201611248741A CN106997407B CN 106997407 B CN106997407 B CN 106997407B CN 201611248741 A CN201611248741 A CN 201611248741A CN 106997407 B CN106997407 B CN 106997407B
- Authority
- CN
- China
- Prior art keywords
- scene
- wind resource
- trend
- wind
- sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Wind Motors (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a wind resource scene reduction method based on trend fitting, which is characterized by comprising the following steps of: analyzing the wind resource scene process lines and performing trend fitting, then performing trend fitting on each wind resource scene process line in the wind resource scene process cluster to obtain a corresponding global trend sequence, then performing scene reduction parameter assignment, and then calculating the distance between the global trend sequences in the wind resource scene process cluster before scene reduction in the current round; and finally, reducing the scenes and updating the scene probability until a scene set and a probability thereof meeting the requirement of the number of reduced scenes are obtained. The similarity between wind resource scenes is judged by automatically refining the trend of the wind resource scene process line, similar scenes are reduced, the purpose of scene reduction is further achieved, random interference of wind resources is avoided, judgment of the similarity of the wind resource scenes is facilitated, the typical scene extraction effect of the wind resources is enhanced, and the method has important significance for wind energy development and utilization.
Description
Technical Field
The invention belongs to the technical field of wind resource characteristic analysis, and particularly relates to a wind resource scene reduction method based on trend fitting.
Background
The research of wind resource scene reduction is mainly used for extracting a typical scene of wind resources, reduces the workload of wind resource analysis, improves the analysis efficiency, and further provides technical support for wind energy development and utilization. The effect of improving the wind resource scene reduction is beneficial to enhancing the recognition capability of the typical scene of the wind resource and improving the technical level of wind resource analysis. At present, wind resource scene reduction technologies at home and abroad directly process wind resource time sequences, identify scenes with short time sequence distances as similar scenes, and delete the similar scenes so as to achieve the purpose of scene reduction. Similar scenes means scenes with similar trends. Wind resources are both trending and stochastic. The interference of the randomness of the wind resources on the calculation of the time sequence distance of the wind resources is not considered in the prior art in the field, and the recognition effect of the similar scene of the wind resources is influenced.
Disclosure of Invention
The technical problem to be solved by the invention is to overcome the defects of the prior art, and provide a wind resource scene reduction method based on trend fitting, so that the scene reduction is carried out according to the trend of wind resources, the random interference of the wind resources is avoided, the typical scene extraction effect of the wind resources is improved, and the technical support is better provided for the development and utilization of wind energy.
In order to solve the technical problems, the invention adopts the following technical scheme:
a wind resource scene reduction method based on trend fitting is characterized by comprising the following steps:
step 1: analyzing a wind resource scene process line according to the relation between wind speed and time in a rectangular coordinate system, setting the set wind resource scene process line after analysis to be composed of N control points in a connected mode, wherein N is larger than or equal to 2, NN wind resource scene process lines form a wind resource scene process cluster, and NN is larger than or equal to 3;
step 2: if N is more than 3, the wind resource scene process line is segmented, and then trend fitting is carried out on specific subintervals of the segmented wind resource scene process line respectively to obtain trend sequences which are respectively corresponding and used as local trends; then fitting the local trend into the global trend of the process line of the wind resource scene; if N is less than or equal to 3, directly performing trend fitting on the wind resource scene process line without performing wind resource scene process line segmentation to obtain a global trend sequence of the wind resource scene process line;
and 3, step 3: performing trend fitting on each wind resource scene process line in the wind resource scene process cluster according to the method in the step 2 to obtain global trend sequences of all corresponding wind resource scenes, and setting NN global trend sequences in total;
And 4, step 4: a scene reduction parameter assignment step, wherein the number of wind resource scenes to be reduced is set to be MM, namely the number of scenes in a wind resource scene set after the scenes are reduced is NN-MM;
step 5, calculating the distance between the wind resource global trend sequences of each wind resource scene process line in the wind resource scene process cluster before scene reduction in the current round;
step 6, reducing scene and updating scene probability:
determining a wind resource scene with minimum probability according to the probability weight distance between the wind resource global trend sequences, deleting the wind resource scene with minimum probability from the set of the probability weight distance and updating the probability of the wind resource scene which is not deleted, so as to obtain a scene set after the scene is reduced in the current round and the probability corresponding to the scene set; and (4) sequentially executing scene reduction, if the number MM of times of executing scene replacement is less than MM, turning to the step 5 to start a new reduction, and otherwise, ending the loop from the step 5 to the step 6 to obtain a scene set meeting the requirement of the number of the scene reduction and the probability thereof.
Further, the process line of analyzing the wind resource scene in step 1 is specifically performed in the following manner:
the wind resource scene process line is obtained in a rectangular coordinate system by taking time t as an abscissa and taking wind speed w as an ordinate according to the process of wind speed; the analysis process comprises the step of analyzing a wind resource scene process line into a process line formed by connecting a plurality of control points, wherein the control points are N in total, the number of the control points is 1,2, …, N from left to right, and the coordinate of the ith control point is marked as (t) i ,w i ) 1,2, ·, N; recording a wind resource scene process line as { t, w }; and setting a total of NN wind resource scene process lines to form a wind resource scene process cluster, and recording the cluster as { TT, WW }, wherein the serial number of the wind resource scene process line is recorded as ii being 1,2, …, NN.
Further, the step 2 is carried out according to the following three steps:
step 2.1, wind resource scene process line segmentation:
if N is less than or equal to 3, the wind resource scene process line segmentation is not needed, and the step 2.3 is directly carried out;
if N is present>And 3, recording the (sg-1) × m +1 th control point to the (sg +1) × m +1 th control point of the wind resource scene process line { t, w } as an sg-th subinterval, wherein m is an integer, and when N is 4, m is 1, N is taken as>4 hours is taken to be in the range ofSG ═ 1,2, …, SG, int (×) denotes rounding "×"; when m is 1, SG is N-3; when m is>At 1 time, ifThenOtherwiseThus, SG sub-regions can be obtained; recording the SG x m +1 control points from the nth control point of { t, w } as SG +1 subintervals; therefore, SG +1 sub-intervals can be obtained, and m +1 points of two adjacent sub-intervals in the obtained sub-intervals are overlapped, namely the rear m +1 point of the front sub-interval is overlapped with the front m +1 point of the rear sub-interval; the front SG subintervals all have 2m +1 control points, and the control point coordinates are recorded as: The SG +1 sub-interval contains N-SG x m control points, and the coordinates of the control points are recorded as
Step 2.2, fitting the local trend of the wind resource:
if N is present>3, respectively performing trend fitting on the SG +1 subintervals obtained in the step 2.1 to obtain a trend sequence corresponding to each subinterval; the control point coordinates of the trend sequence of the top SG subintervals are noted as:SG-1, 2, …, SG; the control point coordinates of the trend sequence of the SG +1 th subinterval are noted as:for the convenience of distinction, call N>3, obtaining a trend sequence as a local trend in the step 2.2;
step 2.3, synthesizing the global trend of the wind resources:
for convenience of distinguishing, the trend obtained in the step 2.3 is called a global trend;
if N is less than or equal to 3, directly performing trend fitting on the wind resource scene process line { t, w } to obtain the global trendThe sequence of potentials t, glo w f and then entering step 3;
if N is present>3, synthesizing a global trend according to the local trend; the specific synthetic process is as follows: taking the first m control points of the local trend of the 1 st subintervalThe first m control points as the trend sequence of the whole wind resource are recorded asTaking the last N- (SG +1) x m-1 control points of the local trend of the SG +1 th subinterval(SG +1) x m +2 control points to Nth control point as a sequence of global trends of wind resources, i.e. the wind resource And (3) carrying out weighting processing on SG +1 subintervals obtained in the step (2.1) with SG overlapping parts according to the influence degree of adjacent subintervals on the fitting trend of the overlapping parts, and carrying out weighting calculation on the local trend of the part according to the following formula to obtain a weighted local trend sequence of the overlapping parts:
wherein: the sg-th overlapped part means that the last m +1 control points of the sg-th subinterval are overlapped with the first m +1 control points of the sg + 1-th subinterval;is the ordinate of the jth control point in the weighted local trend series of the sg-th overlap, j ═ 1,2 1 、λ 2 In order to be the weight coefficient, is the (j + m) th control point ordinate of the local trend sequence of the sg-th sub-interval,is the ordinate of the jth control point in the fitted sequence of the sg +1 th subinterval; let the coordinates of the control points of the weighted local trend series of the sg-th overlap be:SG overlapping parts are weighted to obtain SG weighted local trend sequences, and the coordinates of the last control point of the previous weighted local trend sequence are the same as the coordinates of the first control point of the next weighted local trend sequence; carrying out duplicate removal operation; the specific processing of the duplication elimination is to reserve the coordinate of the last control point of the previous weighted local trend sequence and eliminate the coordinate of the first control point of the next weighted local trend sequence; so de-duplicated, the 1 st weighted local trend sequence contains m +1 control points, i.e. The remaining weighted local trend sequences each contain m control points, i.e.Connecting the SG weighted local trend sequences after de-duplication end to end, and taking the sequences as the (m +1) th control point to the (SG +1) x m +1 th control point of the wind resource global trend sequence, namely
Thus, a wind resource global trend sequence comprising N control points, also denoted as t, glo w f }。
further, step 5 is performed as follows:
recording the wind resource scene set before the current round of reduction as { TT, WW } bef The number of wind resource scenes before the current round of reduction is recorded as KK, the sequence number of the wind resource scenes is recorded as KK, namely KK is 1,2, … and KK, and the probability corresponding to the wind resource scenes is recorded as p kk (ii) a When the step is executed for the 1 st time, the probability that NN wind resource scenes appear is considered to be equal, and is marked as p kk 1/NN; according to { TT, WW } bef Calculating the probability weight distance between every two wind resource global trend sequences by the global trend sequence corresponding to each wind resource scene process line, and recording the probability weight distance as PD kk The formula is as follows:
For the kth scene, a set containing KK-1 distances, denoted as { D kk-jj Will { D } kk-jj Mark the scene sequence number jj corresponding to the minimum value in JL kk For KK scenes, a system containing KK JL's is obtained kk Set of (1) denoted as { JL kk }; for the wind resource scene set before the current round of reduction { TT, WW } bef A set of KK probability weight distances can be obtained, and is marked as { PD kk Will { PD } kk And marking the scene sequence number kk corresponding to the minimum value in the video sequence as PDmin.
Further, step 6 is performed as follows:
the scene with the sequence number PDmin is selected from { TT, WW } bef Deleting; in the { JL kk Find out the element JL of PDmin PDmin Then the probability p of the scene with the sequence number PDmin is used PDmin To number JL PDmin Probability of the scene of (1)Namely, it isThus, a scene set { TT, WW } obtained after the current round of scene reduction can be obtained aft And their corresponding probabilities; the number of times of executing this step is recorded as mm, if mm<And MM, turning to the step 5 to start a new reduction, otherwise, ending the loop from the step 5 to the step 6 to obtain the scene set and the probability thereof meeting the requirement of the number of the reduced scenes.
The wind resource scene reduction technical scheme based on trend fitting provided by the invention judges the similarity between wind resource scenes by automatically refining the trend of the process line of the wind resource scene, provides a new judgment method, has simple and clear result and is simple and easy to implement. In the wind energy resource analysis application, the wind resource scene process and the number of scenes needing to be reduced are used as input, so that the similarity degree between the scenes can be automatically judged, the similar scenes can be reduced, and the purpose of reducing the scenes can be further achieved. Compared with the prior art, the scene similarity identification is carried out by taking the trend of the wind resource as a judgment basis for the first time, so that the random interference of the wind resource is avoided, the method is an important innovation in the technical field, is beneficial to the judgment of the scene similarity of the wind resource, enhances the typical scene extraction effect of the wind resource, has important significance for the development and utilization of wind energy, and has important popularization and use values.
Drawings
FIG. 1 is a schematic view of a sub-interval segmentation of a wind resource scene process line implemented according to the present invention. Fig. 2 is a schematic view of scene composition according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of an initial scene set before reduction according to an embodiment of the present invention.
Fig. 4 is a process of scene reduction using a conventional scene reduction technique, namely, a synchronous back-substitution reduction method (the number of scenes in the figure is the number of reduced scenes).
Fig. 5 is a scene reduction process using the technical solution of the present invention (the number of scenes in the figure is the number of reduced scenes).
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be described below with reference to the embodiments of the present invention and the accompanying drawings.
The invention provides a wind resource scene reduction method based on trend fitting, which comprises the following steps:
The wind resource scene process line is obtained in a rectangular coordinate system by taking time t as an abscissa and taking wind speed w as an ordinate according to the process of wind speed; the analysis process comprises the step of analyzing a wind resource scene process line into a process line formed by connecting a plurality of control points, wherein N control points are arranged, the numbers from left to right are sequentially 1,2, …, N, and the coordinate of the ith control point is recorded as (t) i ,w i ) 1,2, ·, N; and recording the wind resource scene process line as t, w. And setting a total of NN wind resource scene process lines to form a wind resource scene process cluster, and recording the cluster as { TT, WW }, wherein the serial number of the wind resource scene process line is recorded as ii being 1,2, …, NN.
Step 2.1, wind resource scene process line segmentation
If N is less than or equal to 3, the wind resource scene process line segmentation is not needed, and the step 2.3 is directly carried out.
If N is present>3, marking the (sg-1) x m +1 control points to the (sg +1) x m +1 control points of the wind resource scene process line { t, w } as the sg sub-interval, wherein m is an integer and the value range is SG 1,2, …, SG, int (×) indicates rounding to "×". When m is 1, SG is N-3; when m is>At 1 time, ifThenOtherwiseThus, SG sub-intervals can be obtained. Let SG + m +1 control point to nth control point of { t, w } be the SG +1 subinterval. At this point in time,SG +1 subintervals can be obtained, and m +1 points of two adjacent subintervals in the obtained subintervals overlap, that is, the last m +1 points of the preceding subintervals overlap with the first m +1 points of the following subintervals. The front SG subintervals all have 2m +1 control points, and the control point coordinates are recorded as:the SG +1 sub-interval contains N-SG multiplied by m control points, and the coordinates of the control points are recorded as The sub-interval division diagram can be seen in fig. 1, where the sub-interval 1 includes the 1 st control point to the 2m +1 st control point, and the coordinates are expressed asThe subinterval 2 comprises the m +1 th control point to the 3m +1 th control point, and the coordinates are recorded asThe sub-interval sg includes the (sg-1) × m +1 st control point to the (sg +1) × m +1 st control point, and the coordinates are recorded asThe sub-interval SG includes (SG-1) × m +1 control points to (SG +1) × m +1 control points, and the coordinates are recorded asThe sub-interval SG +1 comprises from the SG x m +1 control point to the Nth control point, and the coordinates are recorded asThe specific value of m in this step is set by those skilled in the art.
Step 2.2, wind resource local trend fitting
If N is present>3, respectively performing trend fitting on the SG +1 subintervals obtained in the step 2.1 to obtain corresponding trend sequences; the control point coordinates of the trend sequence of the top SG subintervals are noted as: SG-1, 2, …, SG; the control point coordinates of the trend sequence of the SG +1 th subinterval are noted as:for the convenience of distinction, call N>The trend sequence obtained in this step is a local trend at 3. The trend fitting of this step adopts a polynomial fitting method, and the fitting order is set by those skilled in the art.
Step 2.3, wind resource global trend synthesis
For the convenience of distinction, the trend obtained in this step is called a global trend. If N is less than or equal to 3, directly performing trend fitting on the wind resource scene process line { t, w } to obtain a global trend sequence { t, glo w f And then proceed to step 3. If N is present>And 3, synthesizing a global trend according to the local trend. The specific synthetic process is as follows: taking the first m control points of the local trend of the 1 st subintervalThe first m control points as the global trend sequence of the wind resource are recorded asTaking the last N- (SG +1) x m-1 control points of the local trend of the SG +1 th subinterval(SG +1) x m +2 control points to Nth control point as a sequence of global trends of wind resources, i.e. the wind resourceSG overlapping parts exist in SG +1 subintervals obtained in step 2.1, and the local trend of the overlapping parts is weighted and calculated according to the following formula to obtain a weighted local trend sequence of the overlapping parts.
Wherein: the sg-th overlapped part means that the last m +1 control points of the sg-th subinterval are overlapped with the first m +1 control points of the sg + 1-th subinterval;j ═ 1, 2.. multidot.m +1, λ. 1 、λ 2 In order to be the weight coefficient, is the (j + m) th control point ordinate of the local trend sequence of the sg-th sub-interval,is the ordinate of the jth control point in the fitted sequence of the sg +1 th subinterval. The function of the equation is to perform weighting processing according to the influence degree of the adjacent subintervals on the fitting trend of the overlapping parts of the subintervals.
Let the coordinates of the control points of the weighted local trend series of the sg-th overlap be:and weighting the SG overlapped parts to obtain SG weighted local trend sequences, wherein the coordinate of the last control point of the former weighted local trend sequence is the same as the coordinate of the first control point of the latter weighted local trend sequence. And carrying out the duplicate removal operation. The specific processing of the duplication elimination is to reserve the last control point coordinate of the previous weighted local trend sequence and remove the first control point coordinate of the next weighted local trend sequence. After such deduplication, the 1 st weighted local trend sequence contains m +1 control points, i.e.The remaining weighted local trend sequences each contain m control points, i.e.Connecting the SG weighted local trend sequences after de-duplication end to end, and taking the sequences as the (m +1) th control point to the (SG +1) x m +1 th control point of the wind resource global trend sequence, namely
Thus, a wind resource global trend sequence comprising N control points, also denoted as t, glo w f }. Step 3, global trend sequence of all wind resource scenes
According to the steps 2.1 to 2.3, for each wind resource scene process line in the wind resource scene process cluster, a corresponding wind resource global trend sequence can be obtained, which is marked as t, glo w f } ii And NN in total.
And setting the number of wind resource scenes to be reduced as MM, namely setting the number of scenes in the wind resource scene set after scene reduction as NN-MM.
Step 5, calculating the probability weight distance between the wind resource global trend sequences before the scene reduction of the current round
Recording the wind resource scene set before the current round of reduction as { TT, WW } bef The number of wind resource scenes before the current round of reduction is recorded as KK, the sequence number of the wind resource scenes is recorded as KK, namely KK is 1,2, … and KK, and the probability corresponding to the wind resource scenes is recorded as p kk (ii) a When the step is executed for the 1 st time, the probabilities of occurrence of NN wind resource scenes are regarded as equal and are marked as p kk 1/NN; according to { TT, WW } bef Calculating the probability weight distance between every two wind resource global trend sequences by the global trend sequence corresponding to each wind resource scene process line, and recording the probability weight distance as PD kk The formula is as follows:
For the kth scene, a set containing KK-1 distances, denoted as { D kk-jj Will { D }, will { D kk-jj Mark the scene sequence number jj corresponding to the minimum value in JL kk For KK scenes, a data field containing KK JLs is obtained kk Set of (1) denoted as { JL kk }; for the wind resource scene set before the current round of reduction { TT, WW } bef A set of KK probability weight distances can be obtained, and is marked as { PD kk Will { PD } kk And marking the scene sequence number kk corresponding to the minimum value in the video sequence as PDmin.
And 6: reducing scene and updating scene probabilities
The scene with the sequence number PDmin is selected from { TT, WW } bef Deleting; in the { JL kk Find out the element JL of PDmin PDmin Then the probability p of the scene with the sequence number PDmin is used PDmin To number JL PDmin Probability of a scene of (2)Namely, it isThus, a scene set { TT, WW } obtained after the current round of scene reduction can be obtained aft And its corresponding probability; the number of times of executing this step is recorded as mm, if mm<MM, turning to step 5 to start a new reduction, otherwise, ending the loop from step 5 to step 6 to obtain the scene set and the probability thereof meeting the requirement of the number of the reduced scenes.
Fig. 2 is a schematic diagram of generating a scene process cluster, in which random process lines complying with a standard normal distribution are respectively superimposed by a sinusoidal trend and a step trend to correspond to a synthetic scene process line 1 and a scene process line 2. By superimposing the two trends with 200 process lines that are randomly generated and both obey the standard normal distribution, an initial scene set (reduced initial scene set) containing 400 scene process lines can be obtained, as shown in fig. 3.
FIG. 4 is a process for scene reduction using a prior art scene reduction technique, synchronous back-substitution subtraction. Fig. 5 is a process of scene reduction using the technical solution of the present invention. As can be seen from comparing fig. 4 and fig. 5, the two different trends of sine and staircase are not identified in the scene reduction in the conventional scene reduction technology, but the two different trends of sine and staircase are identified in the technical solution provided by the present invention, so that the effectiveness of the technical solution provided by the present invention is verified.
The invention can be implemented automatically by adopting computer software technology.
According to the embodiment results, different trends are identified by the technical scheme provided by the invention, and the effectiveness of the invention is illustrated. It can be seen that the method can automatically and effectively extract the trend of wind resource scenes and reduce the scenes based on the trend, and provides decision support for wind energy development and utilization.
The method is mainly applied to wind resource scene reduction, and in wind energy resource analysis application, the wind resource scene process and the number of scenes needing to be reduced are used as input, so that the similarity degree between the scenes can be automatically judged, the similar scenes are reduced, and the purpose of scene reduction is further achieved. Compared with the prior art, the invention has the innovation that the similarity between scenes is judged through the trend. In view of this, the present invention and the prior art are applied to scene reduction of a scene process cluster obtained by random superposition of different trends and the same distribution at the same time, so as to verify the rationality of the technical scheme of the present invention.
It should be emphasized that the embodiments described herein are illustrative rather than restrictive, and thus the present invention is not limited to the embodiments described in the detailed description, but other embodiments derived from the technical solutions of the present invention by those skilled in the art are also within the scope of the present invention.
Claims (4)
1. A wind resource scene reduction method based on trend fitting is characterized by comprising the following steps:
step 1: analyzing a wind resource scene process line according to the relation between wind speed and time in a rectangular coordinate system, setting the set wind resource scene process line after analysis to be composed of N control points in a connected mode, wherein N is larger than or equal to 2, NN wind resource scene process lines form a wind resource scene process cluster, and NN is larger than or equal to 3;
step 2: if N is more than 3, the wind resource scene process line is segmented, and then trend fitting is respectively carried out on specific subintervals of the segmented wind resource scene process line to obtain trend sequences which are respectively corresponding and used as local trends; then fitting the local trend into the global trend of the process line of the wind resource scene; if N is less than or equal to 3, directly performing trend fitting on the wind resource scene process line without performing wind resource scene process line segmentation to obtain a global trend sequence of the wind resource scene process line;
the step 2 is distinguished and carried out according to the following three steps:
step 2.1, wind resource scene process line segmentation:
if N is less than or equal to 3, the wind resource scene process line segmentation is not needed, and the step 2.3 is directly carried out;
if N is present>And 3, recording the (sg-1) × m +1 th control point to the (sg +1) × m +1 th control point of the wind resource scene process line { t, w } as an sg-th subinterval, wherein m is an integer, and when N is 4, m is 1, N is taken as >4 time span is in the range ofSG ═ 1,2, …, SG, int (×) denotes rounding "×"; when m is 1, SG is N-3; when m is>At 1 time, ifThen theOtherwiseThus, can obtainUp to SG sub-intervals; recording the SG +1 control point from the (t, w) th control point to the Nth control point as an SG +1 subinterval; so far, SG +1 sub-intervals can be obtained, and m +1 points of two adjacent sub-intervals in the obtained sub-intervals are overlapped, namely the rear m +1 point of the front sub-interval is overlapped with the front m +1 point of the rear sub-interval; the front SG subintervals all have 2m +1 control points, and the coordinates of the control points are recorded asThe SG +1 sub-interval contains N-SG x m control points, and the coordinates of the control points are recorded as
Step 2.2, fitting the local trend of the wind resource:
if N is present>3, respectively carrying out trend fitting on the SG +1 subintervals obtained in the step 2.1 to obtain a trend sequence corresponding to each subinterval; the control point coordinates of the trend sequence of the top SG subintervals are noted as:SG-1, 2, …, SG; the control point coordinates of the trend sequence of the SG +1 th subinterval are noted as:for the convenience of distinction, call N>3, obtaining a trend sequence as a local trend in the step 2.2;
step 2.3, synthesizing the global trend of the wind resources:
for the convenience of distinguishing, the trend obtained in the step 2.3 is called a global trend;
If N is less than or equal to 3, directly performing trend fitting on the wind resource scene process line { t, w } to obtain a global trend sequence { t, glo w f and then entering step 3;
if N is present>3, synthesizing a global trend according to the local trend; the specific synthesis process comprises the following steps: the first m control points of the 1 st subinterval local trend are takenThe first m control points as the global trend sequence of the wind resource are recorded asTaking the last N- (SG +1) x m-1 control points of the local trend of the SG +1 th subintervalFrom the (SG +1) × m +2 control points to the nth control point as the wind resource global trend sequence, that is:
SG +1 subintervals obtained in step 2.1 have SG overlapping parts, weighting processing is carried out according to the influence degree of adjacent subintervals on the fitting trend of the overlapping parts, and the local trend of the overlapping parts is weighted according to the following formula to obtain a weighted local trend sequence of the overlapping parts:
wherein: the sg-th overlapped part means that the last m +1 control points of the sg-th subinterval are overlapped with the first m +1 control points of the sg + 1-th subinterval;for the ordinate of the jth control point in the weighted local trend series of the sg-th overlap, j is 1,2 1 、λ 2 In order to be a weight coefficient of the image, is the sThe (j + m) th control point ordinate of the local trend sequence of g subintervals, Is the ordinate of the jth control point in the fitting sequence of the sg +1 th subinterval; let the coordinates of the control points of the weighted local trend series of the sg-th overlap be:SG overlapping parts are weighted to obtain SG weighted local trend sequences, and the coordinates of the last control point of the previous weighted local trend sequence are the same as the coordinates of the first control point of the next weighted local trend sequence; carrying out duplicate removal operation; the specific processing of the duplication elimination is to reserve the coordinate of the last control point of the previous weighted local trend sequence and eliminate the coordinate of the first control point of the next weighted local trend sequence; so deduplicated, the 1 st weighted local trend sequence contains m +1 control points, i.e.The remaining weighted local trend sequences each contain m control points, i.e.Connecting the SG weighted local trend sequences after de-duplication end to end, and taking the sequences as the (m +1) th control point to the (SG +1) x m +1 th control point of the wind resource global trend sequence, namely
Thus, a wind resource global trend sequence comprising N control points, also denoted as t, glo w f };
and step 3: performing trend fitting on each wind resource scene process line in the wind resource scene process cluster according to the mode in the step 2 to obtain global trend sequences of all corresponding wind resource scenes, and setting NN global trend sequences in total;
And 4, step 4: a scene reduction parameter assignment step, namely setting the number of wind resource scenes to be reduced as MM, namely setting the number of scenes in a wind resource scene set after scene reduction as NN-MM;
step 5, calculating the distance between the wind resource global trend sequences of each wind resource scene process line in the wind resource scene process cluster before scene reduction in the current round;
step 6, reducing the scene and updating the scene probability:
determining a wind resource scene with minimum probability according to probability weight distances among the wind resource global trend sequences, deleting the wind resource scene with minimum probability from a set of probability weight distances and updating the probability of the wind resource scene which is not deleted, so as to obtain a scene set after the scene is reduced in the current round and the probability corresponding to the scene set; and (4) sequentially executing scene reduction, if the number MM of times of executing scene replacement is less than MM, turning to the step 5 to start a new reduction, and otherwise, ending the loop from the step 5 to the step 6 to obtain a scene set meeting the requirement of the number of the scene reduction and the probability thereof.
2. The trend fitting-based wind resource scene reduction method according to claim 1, wherein the step 1 of analyzing the wind resource scene process line is specifically performed by the following steps:
the wind resource scene process line is obtained in a rectangular coordinate system by taking time t as an abscissa and taking wind speed w as an ordinate according to the process of wind speed; the analysis process comprises the step of analyzing a wind resource scene process line into a process line formed by connecting a plurality of control points, wherein N control points are arranged, the numbers from left to right are sequentially 1,2, …, N, and the coordinate of the ith control point is recorded as (t) i ,w i ) 1,2, ·, N; recording a wind resource scene process line as { t, w }; and setting a total of NN wind resource scene process lines to form a wind resource scene process cluster, and recording the cluster as { TT, WW }, wherein the serial number of the wind resource scene process line is recorded as ii being 1,2, …, NN.
3. The trend fitting-based wind resource scene reduction method according to claim 2, wherein the step 5 is performed as follows:
recording the wind resource scene set before the current round of reduction as { TT, WW } bef The number of wind resource scenes before the current round of reduction is recorded as KK, the sequence number of the wind resource scenes is recorded as KK, namely KK is 1,2, … and KK, and the probability corresponding to the wind resource scenes is recorded as p kk (ii) a When the step is executed for the 1 st time, the probability that NN wind resource scenes appear is considered to be equal, and is marked as p kk 1/NN; according to { TT, WW } bef Calculating the probability weight distance between every two wind resource global trend sequences by the global trend sequence corresponding to each wind resource scene process line, and recording the probability weight distance as PD kk The formula is as follows:
For the kth scene, a set containing KK-1 distances can be obtained, denoted { D kk-jj Will { D } kk-jj The scene sequence number jj corresponding to the minimum value in the JL is marked as JL kk For KK scenes, a data field containing KK JLs is obtained kk Set of (1) denoted as { JL kk }; for the wind resource scene set before the current round of reduction { TT, WW } bef A set of KK probability weight distances can be obtained, denoted as { PD kk Will { PD }, will { PD kk And marking the scene serial number kk corresponding to the minimum value in the images as PDmin.
4. The trend fitting-based wind resource scene reduction method according to claim 3, wherein the step 6 is performed as follows:
the scene with the sequence number PDmin is selected from { TT, WW } bef Deleting; in the { JL kk Find out the element JL of PDmin PDmin Then the probability p of the scene with the sequence number PDmin is used PDmin To number JL PDmin Probability of the scene of (1)Namely, it isThus, a scene set { TT, WW } obtained after the current round of scene reduction can be obtained aft And its corresponding probability; the number of times of executing this step is recorded as mm, if mm<MM, turning to step 5 to start a new reduction, otherwise, ending the loop from step 5 to step 6 to obtain a scene set meeting the requirement of the number of reduced scenes and the probability thereof.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611248741.7A CN106997407B (en) | 2016-12-29 | 2016-12-29 | Wind resource scene reduction method based on trend fitting |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611248741.7A CN106997407B (en) | 2016-12-29 | 2016-12-29 | Wind resource scene reduction method based on trend fitting |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106997407A CN106997407A (en) | 2017-08-01 |
CN106997407B true CN106997407B (en) | 2022-08-16 |
Family
ID=59431381
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611248741.7A Active CN106997407B (en) | 2016-12-29 | 2016-12-29 | Wind resource scene reduction method based on trend fitting |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106997407B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110321799B (en) * | 2019-06-04 | 2022-11-18 | 武汉大学 | Scene number selection method based on SBR and average inter-class distance |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104156543A (en) * | 2014-08-28 | 2014-11-19 | 国家电网公司 | Major power grid failure successive on-off disturbance setting method for whole-process simulation |
CN104268436A (en) * | 2014-10-20 | 2015-01-07 | 武汉大学 | Trend fitting based wind power intraday fluctuation continuous period recognition method and system |
CN105631520A (en) * | 2015-07-16 | 2016-06-01 | 长沙理工大学 | Novel modeling method for extracting random and fuzzy uncertainty characteristics of wind speed |
CN106230028A (en) * | 2016-09-08 | 2016-12-14 | 安徽电气工程职业技术学院 | A kind of Multipurpose Optimal Method of wind-powered electricity generation water-storage association system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103903246A (en) * | 2012-12-26 | 2014-07-02 | 株式会社理光 | Object detection method and device |
-
2016
- 2016-12-29 CN CN201611248741.7A patent/CN106997407B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104156543A (en) * | 2014-08-28 | 2014-11-19 | 国家电网公司 | Major power grid failure successive on-off disturbance setting method for whole-process simulation |
CN104268436A (en) * | 2014-10-20 | 2015-01-07 | 武汉大学 | Trend fitting based wind power intraday fluctuation continuous period recognition method and system |
CN105631520A (en) * | 2015-07-16 | 2016-06-01 | 长沙理工大学 | Novel modeling method for extracting random and fuzzy uncertainty characteristics of wind speed |
CN106230028A (en) * | 2016-09-08 | 2016-12-14 | 安徽电气工程职业技术学院 | A kind of Multipurpose Optimal Method of wind-powered electricity generation water-storage association system |
Non-Patent Citations (2)
Title |
---|
Influence of Sub-Daily Variation on Multi-Fractal Detrended Fluctuation Analysis of Wind Speed Time Series;Xianxun Wang,et al.;《PloS ONE》;20160107;第11卷(第1期);第1-14页 * |
基于量维和形维的风水联合出力波动量化评估;王现勋;《华中科技大学学报》;20160531;第44卷(第5期);第103-107、132页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106997407A (en) | 2017-08-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107591800B (en) | Method for predicting running state of power distribution network with distributed power supply based on scene analysis | |
JP5506785B2 (en) | Fingerprint representation using gradient histogram | |
CN109189991A (en) | Repeat video frequency identifying method, device, terminal and computer readable storage medium | |
EP3189469B1 (en) | A method for selecting frames from video sequences based on incremental improvement | |
CN106156755A (en) | Similarity calculating method in a kind of recognition of face and system | |
US9177020B2 (en) | Gathering index statistics using sampling | |
CN109118420B (en) | Watermark identification model establishing and identifying method, device, medium and electronic equipment | |
CN109978006B (en) | Face image clustering method and device | |
CN112270290B (en) | Dynamic clustering method and device for face images, electronic equipment and storage medium | |
KR102116791B1 (en) | Apparatus, method and computer readable recording medium for detecting moving object using depth map | |
CN106997407B (en) | Wind resource scene reduction method based on trend fitting | |
CN115965657A (en) | Target tracking method, electronic device, storage medium, and vehicle | |
TWI717655B (en) | Feature determination apparatus and method adapted to multiple object sizes | |
EP4044014A1 (en) | Data reduction method and apparatus, computing device, and storage medium | |
CN110321799B (en) | Scene number selection method based on SBR and average inter-class distance | |
CN113963197A (en) | Image recognition method and device, electronic equipment and readable storage medium | |
JP7299334B2 (en) | Chunking method and apparatus | |
CN112257689A (en) | Training and recognition method of face recognition model, storage medium and related equipment | |
CN111552963B (en) | Malicious software classification method based on structure entropy sequence | |
CN103678545A (en) | Network resource clustering method and device | |
KR101671227B1 (en) | Apparatus for detecting duplication and similarity of documents, method thereof and computer recordable medium storing the method | |
CN111125193B (en) | Method, device, equipment and storage medium for identifying abnormal multimedia comments | |
US20150067303A1 (en) | Input data aggregation processing apparatus, system and method | |
CN114816838A (en) | Method, device, medium and computer equipment for improving data recovery efficiency | |
CN113032610A (en) | Archive management method, device, equipment and computer readable storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |