CN106997407A - Wind-resources scene reduction method based on trend fitting - Google Patents

Abstract

The invention discloses a wind resource scene reduction method based on trend fitting, which is characterized by comprising the following steps: analyzing the wind resource scene process line and performing trend fitting, and then for each wind resource scene process in the wind resource scene process cluster Trend fitting is carried out on the average line to obtain the corresponding global trend sequence, and then the scene reduction parameter assignment is performed, and then the distance between each global trend sequence in the wind resource scene process cluster before the current round of scene reduction is calculated; finally, the reduction scene and the update scene are performed Probabilities until the set of scenes and their probabilities that meet the requirements of reducing the number of scenes are obtained. By automatically extracting the trend of the wind resource scene process line to judge the similarity between the wind resource scenes, reduce similar scenes, and then achieve the purpose of scene reduction, avoid the random interference of wind resources, which is beneficial to the judgment of the similarity of wind resource scenes, and enhance The extraction effect of typical scenes of wind resources is of great significance for the development and utilization of wind energy.

Description

Wind resource scene reduction method based on trend fitting

technical field

The invention belongs to the technical field of wind resource characteristic analysis, in particular to a method for reducing wind resource scenarios based on trend fitting.

Background technique

The research on wind resource scene reduction is mainly used to extract typical wind resource scenes, reduce the workload of wind resource analysis, improve analysis efficiency, and provide technical support for wind energy development and utilization. Improving the reduction effect of wind resource scenarios will help to strengthen the ability to identify typical scenarios of wind resources and improve the technical level of wind resource analysis. At present, wind resource scene reduction technologies both at home and abroad directly process the time series of wind resources, identify scenes with close time series distances as similar scenes, and delete similar scenes to achieve the purpose of scene reduction. Similar scenarios mean scenarios with similar trends. Wind resources are both trendy and random. The existing technologies in this field have not considered the interference of the randomness of wind resources on the distance calculation of wind resource time series, which affects the recognition effect of similar scenes of wind resources.

Contents of the invention

The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a wind resource scene reduction method based on trend fitting, which can reduce the scene according to the trend of wind resources, avoid random interference of wind resources, and improve the wind resource typicality. The effect of scene extraction can better provide technical support for the development and utilization of wind energy.

In order to solve the problems of the technologies described above, the present invention adopts the following technical solutions:

A method for reducing wind resource scenarios based on trend fitting, characterized in that it comprises the following steps:

Step 1: Analyze the wind resource scenario process line according to the relationship between wind speed and time in the Cartesian coordinate system, and set the wind resource scenario process line after analysis to be composed of N control point connections, N≥2, NN wind resource scenarios The process lines form a wind resource scenario process cluster, NN≥3;

Step 2: If N>3, segment the wind resource scene process line, and then perform trend fitting on the specific sub-intervals of the wind resource scene process line after the segmentation, and obtain the corresponding trend sequences as local trends; then Fit the local trend to the global trend of the wind resource scenario process line; if N≤3, there is no need to segment the wind resource scenario process line, and directly carry out trend fitting on the wind resource scenario process line to obtain the wind resource scenario process line global trend sequence;

Step 3: For each wind resource scenario process line in the wind resource scenario process cluster, carry out trend fitting according to the method of step 2, obtain the corresponding global trend sequences of all wind resource scenarios, and set a total of NN global trend sequences;

Step 4: Scenario reduction parameter assignment step, set the number of wind resource scenarios to be reduced to MM, that is, the number of scenarios in the wind resource scenario set after scenario reduction is NN-MM;

Step 5, calculate the distance between the global trend sequences of wind resources of each wind resource scenario process line in the wind resource scenario process cluster before the current round of scenario reduction;

Step 6, reduce the scene and update the probability of the scene:

Determine the minimum probability wind resource scenario according to the probability weight distance between the global wind resource trend sequences, delete the minimum probability wind resource scenario from the probability weight distance set and update the probability of the undeleted wind resource scenario, and get the current round The set of scenes after reducing the scenes and their corresponding probabilities; perform scene reduction in sequence, if the number of scene replacements mm<MM, then go to step 5 to start a new round of reduction, otherwise end the cycle from step 5 to step 6, and it is satisfied Scenario collections and their probabilities required to reduce the number of scenarios.

Further, step 1 analyzes the wind resource scenario process line specifically as follows:

The wind resource scene process line is obtained in the Cartesian coordinate system according to the process of wind speed, with time t as the abscissa and wind speed w as the vertical coordinate; the analysis process includes analyzing the wind resource scene process line as being connected by several control points Assuming that there are N control points in total, numbered from left to right are 1, 2, ..., N, and the coordinates of the i-th control point are marked as (t _i , w _i ), i=1, 2, ..., N; record the wind resource scenario process line as {t,w}; suppose there are NN wind resource scenario process lines to form a wind resource scenario process cluster, which is denoted as {TT,WW}, where the wind resource scenario process line number is denoted as ii=1,2,...,NN.

Further, step 2 is carried out according to the following three steps:

Step 2.1, wind resource scene process line segmentation:

If N≤3, there is no need to divide the wind resource scene process line, and go directly to step 2.3;

If N>3, record the (sg-1)×m+1th control point to the (sg+1)×m+1th control point of the wind resource scenario process line {t,w} as the sg-th subinterval , where m is an integer, when N=4, m takes 1, and when N>4, the value range is sg=1,2,...,SG, int(*) means to round "*"; when m=1, SG=N-3; when m>1, if but otherwise So far, SG subintervals can be obtained; the SG×m+1th control point to the Nth control point of {t,w} is recorded as the SG+1th subinterval; so far, SG+1 subintervals can be obtained, and There are m+1 points overlapping between two adjacent sub-intervals in the obtained sub-interval, that is, the last m+1 points of the previous sub-interval overlap with the first m+1 points of the subsequent sub-interval; the first SG sub-intervals all have 2m+1 control points, whose coordinates are marked as: The SG+1th subinterval contains N-SG×m control points, and the coordinates of the control points are marked as

Step 2.2, local trend fitting of wind resources:

If N>3, carry out trend fitting on the SG+1 subintervals obtained in step 2.1 respectively, and obtain the trend sequence corresponding to each subinterval; the control point coordinates of the trend sequence of the first SG subintervals are marked as: sg=1,2,...,SG; the control point coordinates of the trend sequence of the SG+1th subinterval are marked as: For ease of distinction, when N>3, the trend sequence obtained in step 2.2 is called a local trend;

Step 2.3, wind resource global trend synthesis:

For ease of distinction, the trend obtained in step 2.3 is called the global trend;

If N≤3, directly perform trend fitting on the wind resource scenario process line {t,w} to obtain its global trend sequence {t, ^glow w ^f }, and then proceed to step 3;

If N>3, the global trend must be synthesized according to the local trend; the specific synthesis process is: take the first m control points of the local trend in the first subinterval As the first m control points of the global trend sequence of wind resources, denoted as Take the last N-(SG+1)×m-1 control points of the local trend of the SG+1th subinterval As the global trend sequence of wind resources from the (SG+1)×m+2th control point to the Nth control point, that is The SG+1 subintervals obtained from step 2.1 have SG overlapping parts, and weighting is performed according to the influence of adjacent subintervals on the fitting trend of their overlapping parts, and the local trend of this part is weighted according to the following formula to obtain Weighted local trend series for overlapping parts:

Among them: the sg-th overlap refers to the overlap between the last m+1 control points of the sg-th subinterval and 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 sequence of the sgth overlapping part, j=1,2,...,m+1, λ ₁ and λ ₂ are weight coefficients, is the ordinate of the j+mth control point of the local trend sequence of the sgth subinterval, is the ordinate of the jth control point in the fitting sequence of the sg+1th subinterval; the coordinates of the control point of the weighted local trend sequence of the sgth overlapping part are marked as: After weighting the SG overlapping parts, SG weighted local trend sequences can be obtained, 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 latter weighted local trend sequence; Repeat operation; the specific processing of deduplication is to keep the coordinates of the last control point of the previous weighted local trend sequence, and remove the coordinates of the first control point of the subsequent weighted local trend sequence; after such deduplication, the first weighted local trend sequence The trend sequence contains m+1 control points, namely The remaining weighted local trend sequences all contain m control points, namely Connect the SG weighted local trend sequences after deduplication end to end, and take them as the m+1th control point to the (SG+1)×m+1th control point of the global trend sequence of wind resources, that is

So far, the global trend sequence of wind resources including N control points can be obtained, which is also denoted as {t, ^{glow f} }.

Further, step 5 is carried out as follows:

Record the set of wind resource scenes before this round of reduction as {TT,WW} ^bef , the number of wind resource scenes before this round of reduction as KK, and the sequence number of wind resource scenes as kk, that is, kk=1,2,… , KK, the probability corresponding to the wind resource scene is recorded as p _kk ; when this step is executed for the first time, KK=NN, it is considered that the probability of occurrence of NN wind resource scenes is equal, and it is recorded as p _kk =1/NN; according to { TT,WW} ^bef in the global trend sequence corresponding to each wind resource scenario process line to calculate the probability weight distance between two global trend sequences of wind resources, denoted as PD _kk , the formula is as follows:

in kk≠jj, kk=1,2,...,KK, jj=1,2,...,KK;

For the kkth scene, a set containing KK-1 distances can be obtained, denoted as {D _kk-jj }, and the scene number jj corresponding to the minimum value in {D _kk-jj } is marked as JL _kk , for KK scenes A set containing KK JL _kk can be obtained, denoted as {JL _kk }; for the wind resource scene set {TT,WW} ^bef before this round of reduction, a set containing KK probability weight distances can be obtained, denoted as {PD _kk }, mark the scene number kk corresponding to the minimum value in {PD _kk } as PDmin.

Further, step 6 is carried out as follows:

Delete the scene with the serial number PDmin from {TT,WW} ^bef ; find the PDminth element JL _PDmin in {JL _kk }, and then add the probability p _PDmin of the scene with the serial number PDmin to the JL _PDmin Scenario Probability which is So far, the scene set {TT,WW} ^aft and its corresponding probability after the current round of reduced scenes can be obtained; record the number of executions of this step as mm, if mm<MM, go to step 5 to start a new round of reduction, Otherwise, end the cycle from step 5 to step 6, and obtain the scene set and its probability that meet the requirement of reducing the number of scenes.

The trend-fitting-based wind resource scene reduction technical solution provided by the present invention judges the similarity between wind resource scenes by automatically extracting the trend of the wind resource scene process line, and provides a new judgment method. The result is simple and clear, and the implementation Simple and easy. In the application of wind energy resource analysis, the process of wind resource scenarios and the number of scenarios to be reduced can be used as input to automatically judge the similarity between scenarios, reduce similar scenarios, and achieve the purpose of scenario reduction. Compared with the existing technology, it is proposed for the first time to use the trend of wind resources as the basis for the identification of similar scenes to avoid the random interference of wind resources. The extraction effect of typical scenes is of great significance to the development and utilization of wind energy, and has important promotion and use value.

Description of drawings

Fig. 1 is a schematic diagram of sub-interval division of a wind resource scenario process line implemented according to the present invention. Fig. 2 is a schematic diagram of scene synthesis according to the 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 the process of scene reduction using the existing scene reduction technology - synchronous back generation reduction method (the number of scenes in the figure is the number of reduced scenes).

Fig. 5 is the scene reduction process using the technical solution of the present invention according to the embodiment of the present invention (the number of scenes in the figure is the number of scenes after reduction).

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be introduced below in conjunction with the embodiments of the present invention and the accompanying drawings.

The present invention provides a wind resource scenario reduction method based on trend fitting, comprising the following steps:

Step 1, analyze the wind resource scene process line

The wind resource scene process line is obtained in the Cartesian coordinate system according to the process of wind speed, with time t as the abscissa and wind speed w as the vertical coordinate; the analysis process includes analyzing the wind resource scene process line as being connected by several control points Assuming that there are N control points in total, numbered from left to right are 1, 2, ..., N, and the coordinates of the i-th control point are marked as (t _i , w _i ), i=1, 2, ..., N; record the wind resource scenario process line as {t,w}. It is assumed that there are NN wind resource scenario process lines in total to form a wind resource scenario process cluster, denoted as {TT,WW}, and the numbers of the wind resource scenario process lines are recorded as ii=1,2,...,NN.

Step 2.1, wind resource scene process line segmentation

If N≤3, there is no need to divide the wind resource scene process line, and go directly to step 2.3.

If N>3, record the (sg-1)×m+1th control point to the (sg+1)×m+1th control point of the wind resource scenario process line {t,w} as the sg-th subinterval , where m is an integer, and its value range is sg=1,2,...,SG, int(*) means to round "*". When m=1, SG=N-3; when m>1, if but otherwise So far, SG subintervals can be obtained. The SG×m+1th control point to the Nth control point of {t,w} is recorded as the SG+1th subinterval. So far, SG+1 subintervals can be obtained, and m+1 points overlap between two adjacent subintervals in the obtained subintervals, that is, the last m+1 points of the previous subinterval and the first m+1 points of the subsequent subinterval 1 point overlaps. There are 2m+1 control points in the first SG subintervals, and the coordinates of the control points are marked as: The SG+1th subinterval contains N-SG×m control points, and the coordinates of the control points are marked as See Figure 1 for the schematic diagram of sub-interval division. Sub-interval 1 includes the first control point to the 2m+1th control point, and the coordinates are marked as Subinterval 2 includes the m+1th control point to the 3m+1th control point, and the coordinates are marked as The subinterval sg includes the (sg-1)×m+1th control point to the (sg+1)×m+1th control point, and the coordinates are marked as The subinterval SG includes the (SG-1)×m+1th control point to the (SG+1)×m+1th control point, and the coordinates are marked as The subinterval SG+1 includes the SG×m+1th control point to the Nth control point, and the coordinates are marked as The 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>3, carry out trend fitting on the SG+1 subintervals obtained in step 2.1 to obtain the corresponding trend sequence; the control point coordinates of the trend sequence of the first SG subintervals are marked as: sg=1,2,...,SG; the control point coordinates of the trend sequence of the SG+1th subinterval are marked as: For ease of distinction, when N>3, the trend sequence obtained in this step is called a local trend. The trend fitting in 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 ease of distinction, the trend obtained in this step is called the global trend. If N≤3, directly perform trend fitting on the wind resource scenario process line {t,w} to obtain its global trend sequence {t, ^glow w ^f }, and then proceed to step 3. If N>3, the global trend must be synthesized according to the local trend. The specific synthesis process is: take the first m control points of the local trend of the first subinterval As the first m control points of the global trend sequence of wind resources, denoted as Take the last N-(SG+1)×m-1 control points of the local trend of the SG+1th subinterval As the global trend sequence of wind resources from the (SG+1)×m+2th control point to the Nth control point, that is The SG+1 subintervals obtained from step 2.1 have SG overlapping parts, and the local trends of this part are weighted according to the following formula to obtain the weighted local trend sequence of the overlapping parts.

Among them: the sg-th overlap refers to the overlap between the last m+1 control points of the sg-th subinterval and 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 sequence of the sgth overlapping part, j=1,2,...,m+1, λ ₁ and λ ₂ are weight coefficients, is the ordinate of the j+mth control point of the local trend sequence of the sgth subinterval, It is the ordinate of the jth control point in the fitting sequence of the sg+1th subinterval. The function of this formula is to carry out weighting according to the degree of influence of adjacent subintervals on the fitting trend of their overlapping parts.

Label the control point coordinates of the weighted local trend sequence for the sg-th overlap as: SG weighted local trend sequences can be obtained after SG overlapping parts are weighted, and the coordinates of the last control point of the previous weighted local trend sequence are the same as the first control point coordinates of the latter weighted local trend sequence. Perform deduplication. The specific processing of deduplication is to keep the coordinates of the last control point of the previous weighted local trend sequence, and remove the coordinates of the first control point of the subsequent weighted local trend sequence. After deduplication, the first weighted local trend sequence contains m+1 control points, namely The remaining weighted local trend sequences all contain m control points, namely Connect the SG weighted local trend sequences after deduplication end to end, and take them as the m+1th control point to the (SG+1)×m+1th control point of the global trend sequence of wind resources, that is

So far, the global trend sequence of wind resources including N control points can be obtained, which is also denoted as {t, ^{glow f} }. Step 3, Global Trend Sequence for All Wind Resource Scenarios

According to steps 2.1 to 2.3, for each wind resource scenario process line in the wind resource scenario process cluster, the corresponding global trend sequence of wind resources can be obtained, denoted as {t, ^{glow f} } ⁱⁱ , NN in total.

Step 4, scene reduction parameter assignment

Set the number of wind resource scenes to be reduced as MM, that is, the number of scenes in the set of wind resource scenes after scene reduction is NN-MM.

Step 5. Calculate the probability weight distance between the global trend sequences of wind resources before the current round of scene reduction

Record the set of wind resource scenes before this round of reduction as {TT,WW} ^bef , the number of wind resource scenes before this round of reduction as KK, and the sequence number of wind resource scenes as kk, that is, kk=1,2,… , KK, the probability corresponding to the wind resource scene is recorded as p _kk ; when this step is executed for the first time, KK=NN, it is considered that the probability of occurrence of NN wind resource scenes is equal, and it is recorded as p _kk =1/NN; according to { TT,WW} ^bef in the global trend sequence corresponding to each wind resource scenario process line to calculate the probability weight distance between two global trend sequences of wind resources, denoted as PD _kk , the formula is as follows:

in kk≠jj, kk=1,2,...,KK, jj=1,2,...,KK;

For the kkth scene, a set containing KK-1 distances can be obtained, denoted as {D _kk-jj }, and the scene number jj corresponding to the minimum value in {D _kk-jj } is marked as JL _kk , for KK scenes A set containing KK JL _kk can be obtained, denoted as {JL _kk }; for the wind resource scene set {TT,WW} ^bef before this round of reduction, a set containing KK probability weight distances can be obtained, denoted as {PD _kk }, mark the scene number kk corresponding to the minimum value in {PD _kk } as PDmin.

Step 6: Reduce the scene and update the probability of the scene

Delete the scene with the serial number PDmin from {TT,WW} ^bef ; find the PDminth element JL _PDmin in {JL _kk }, and then add the probability p _PDmin of the scene with the serial number PDmin to the JL _PDmin Scenario Probability which is So far, the scene set {TT,WW} ^aft and its corresponding probability after the current round of reduced scenes can be obtained; record the number of executions of this step as mm, if mm<MM, go to step 5 to start a new round of reduction, Otherwise, end the cycle from step 5 to step 6, and obtain the scene set and its probability that meet the requirement of reducing the number of scenes.

Figure 2 is a schematic diagram of the generation of scene process clusters. The sinusoidal trend and the step trend are respectively superimposed on random process lines that obey the standard normal distribution to correspond to the synthetic scene process line 1 and scene process line 2. By superimposing the above two trends with 200 process lines randomly generated and subject to standard normal distribution, an initial scene set containing 400 scene process lines (initial scene set before reduction) can be obtained, as shown in Figure 3.

FIG. 4 shows the process of scene reduction using the existing scene reduction technology—the synchronous back-generation reduction method. Fig. 5 is a process of scene reduction using the technical solution of the present invention. Comparing Figure 4 and Figure 5, it can be seen that the existing scene reduction technology does not recognize two different trends of sinusoidal and ladder in scene reduction, but the technical solution proposed by the present invention recognizes two different trends of sinusoidal and ladder, The effectiveness of the technical solution proposed in the present invention has been verified.

When the present invention is specifically implemented, computer software technology can be used to realize automatic operation.

As can be seen from the results of the examples, the technical solutions proposed in the present invention have identified different trends, which demonstrates the effectiveness of the present invention. It can be seen that the present invention can automatically and effectively extract the trend of the wind resource scene and reduce the scene based on it, so as to provide decision support for the development and utilization of wind energy.

The present invention is mainly applied to the reduction of wind resource scenarios. In the application of wind energy resource analysis, the wind resource scenario process and the number of scenarios to be reduced are used as input to automatically judge the similarity between the scenarios, reduce similar scenarios, and then achieve the scenario Reduced purpose. Compared with the existing related technologies, the innovation of the present invention lies in judging the similarity between scenes through the trend. In view of this, applying the present invention and the existing technology to the scene reduction of the scene process cluster obtained by the random superposition of different trends and the same distribution can be used to verify the rationality of the technical solution of the present invention.

It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive, so the present invention is not limited to the embodiments described in the specific implementation, and those skilled in the art according to the technical solutions of the present invention Other obtained implementation modes also belong to the protection scope of the present invention.

Claims (5)

1. a kind of wind-resources scene reduction method based on trend fitting, it is characterised in that comprise the following steps：

Step 1：According to wind speed and the relation of time in rectangular coordinate system, wind-resources scene graph is parsed, after setting parsing By N number of control point, connection is constituted setting wind-resources scene graph, N >=2, NN wind-resources scene graph composition wind-resources Scape process cluster, NN >=3；

Step 2：If N>3, wind-resources scene hydrograph separation is carried out, afterwards to the spy of place wind-resources scene graph after segmentation Stator interval carries out trend fitting respectively, obtains the respective corresponding trend sequence as local trend；Afterwards again will be local Global trend of the trend fitting into place wind-resources scene graph；If N≤3, without carrying out wind-resources scene hydrograph separation, Trend fitting directly is carried out to wind-resources scene graph, the global trend sequence of wind-resources scene graph where obtaining；

Step 3：Trend is carried out in the way of step 2 for each wind-resources scene graph in wind-resources scene process cluster Fitting, obtains the global trend sequence of corresponding all wind-resources scenes, shared NN global trend sequence of setting；

Step 4：Scene reduces parameter assignment step, sets the wind-resources scene number for intending reduction as MM, i.e., after scene reduction The number of wind-resources scene set Scene is NN-MM；

Step 5, the wind-resources for calculating each wind-resources scene graph in wind-resources scene process cluster before the reduction of epicycle scene are complete The distance between office's trend sequence；

Step 6, reduction scene and more new scene probability：

The wind-resources scene of minimum probability is determined according to the probability right distance between the global trend sequence of wind-resources, by the minimum The wind-resources scene of probability is deleted from the set of probability right distance and updates the probability of not deleted wind-resources scene, is obtained Scene set and its corresponding probability after epicycle reduction scene；Scene reduction is performed successively, such as performs the number of times that scene is replaced mm<MM, then go to step 5 and start new round reduction, and otherwise end step 5 is met scene reduction to the circulation of step 6 Scene set and its probability that number is required.

2. the wind-resources scene reduction method according to claim 1 based on trend fitting, it is characterised in that step 1 is parsed Wind-resources scene graph step specific as follows is carried out：

The wind-resources scene graph is in rectangular coordinate system, according to the process of wind speed, using time t as abscissa, with wind Fast w obtains for ordinate；Resolving includes resolving to wind-resources scene graph to be connected by some control points constituting, if altogether There is N number of control point, numbering is followed successively by 1,2 from left to right ..., N, i-th of control point coordinates is designated as (t_i,w_i), i=1,2 ..., N；Wind-resources scene graph is designated as { t, w }；If having NN wind-resources scene graph, wind-resources scene process is constituted Cluster, is designated as { TT, WW }, and wherein wind-resources scene graph numbering is designated as ii=1,2 ..., NN.

3. the wind-resources scene reduction method according to claim 2 based on trend fitting, it is characterised in that step 2 is by such as Lower three steps difference is carried out：

Step 2.1, wind-resources scene hydrograph separation：

If N≤3, without carrying out wind-resources scene hydrograph separation, step 2.3 is directly entered；

If N>3, by (sg-1) × m+1 control point to (sg+1) of wind-resources scene graph { t, w } × m+1 control Point is designated as the sg subinterval, and wherein m is integer, and during N=4, m takes 1, N>Span is when 4Sg=1, 2 ..., SG, int (*) represent to round " * "；As m=1, S G=N-3；Work as m>When 1, ifThenOtherwiseSo far, It can obtain SG subinterval；The SG × m+1 control point to the n-th control point of { t, w } is designated as the SG+1 subinterval；Extremely This, can obtain between two subintervals adjacent in SG+1 subinterval, and gained subinterval has m+1 point overlapping, i.e., before The rear m+1 point in subinterval is overlapping with the preceding m+1 point in subinterval below；There is 2m+1 control point in preceding SG subinterval, its Control point coordinates is designated as：Contain N-SG × m control in the SG+1 subinterval Point, it controls point coordinates to be designated as

Step 2.2, wind-resources local trend is fitted：

If N>3, SG+1 subinterval obtained by step 2.1 is subjected to trend fitting respectively, each subinterval is obtained corresponding Trend sequence；The control point coordinates of the trend sequence in preceding SG subinterval is designated as：Sg=1,2 ..., SG；The trend sequence in the SG+1 subinterval Control point coordinates be designated as：For ease of distinguishing, Title works as N>Trend sequence is obtained for local trend in this step 2.2 when 3；

Step 2.3, the global trend synthesis of wind-resources：

For ease of distinguishing, the trend that this step 2.3 is obtained is called global trend；

If N≤3, trend fitting directly is carried out to wind-resources scene graph { t, w }, obtain its global trend sequence t,^glow^f, Subsequently into step 3；

If N>When 3, global trend must be synthesized according to local trend；Specifically building-up process is：Take the 1st subinterval local trend Preceding m control pointsAs the preceding m control point of the global trend sequence of wind-resources, It is designated asTake rear N- (SG+1) × m-1 of the SG+1 subinterval local trend Control pointIt is used as the global trend sequence of wind-resources (SG+1) × m+2 control point to n-th control point, i.e.,The SG+1 subinterval obtained by step 2.1 There is SG lap, according to being weighted processing to the influence degree of its lap fitted trend between adjacent subarea, The local trend of the part is weighted as the following formula, the weighting local trend sequence of lap is obtained：

$w_j^{sg,f}{}_{\mathrm{\&lambda;},loc}={\mathrm{\&lambda;}}_1\&times;w_{j+m}^{sg,f}{}_{loc}+{\mathrm{\&lambda;}}_2\&times;w_j^{sg+1,f}{}_{loc};$

Wherein：The sg lap refers to the rear m+1 control point in the sg subinterval and the preceding m+1 in the sg+1 subinterval Individual control point is overlapping；For the vertical seat at j-th of control point in the weighting local trend sequence of the sg lap Mark, j=1,2 ..., m+1, λ₁、λ₂For weight coefficient, For the office in the sg subinterval + m control point ordinates of jth of portion's trend sequence,For j-th of control in the fitting sequence in the sg+1 subinterval Point ordinate；The control point coordinates of the weighting local trend sequence of the sg lap is designated as：SG can be obtained after processing is weighted to SG lap Individual weighting local trend sequence, and last control point coordinates of previous weighting local trend sequence and latter weighting office First control point coordinates of portion's trend sequence is identical；Carry out deduplication operation；The specific of duplicate removal is processed as previous weighting office Last control point coordinates of portion's trend sequence retains, and removes first control point of its latter weighting local trend sequence Coordinate；After such duplicate removal, the 1st weighting local trend sequence includes m+1 control point, i.e.,Remaining weighting local trend sequence includes m control point, i.e.,SG weighting local trend sequence after duplicate removal is joined end to end, takes it to be provided as wind The m+1 control point to (SG+1) × m+1 control point of source overall situation trend sequence, i.e.,

So far, the global trend sequence of the wind-resources comprising N number of control point can be obtained, be also designated as t,^glow^f}。

4. the wind-resources scene reduction method according to claim 3 based on trend fitting, it is characterised in that step 5 is by such as Lower step is carried out：

Wind-resources scene set before epicycle is reduced is designated as { TT, WW }^bef, the number of wind-resources scene is designated as before epicycle reduction KK, the sequence number of wind-resources scene is designated as kk, i.e. kk=1, and 2 ..., KK, the probability of correspondence wind-resources scene are designated as p_kk；When the 1st time When performing this step, KK=NN, the probability for being regarded as NN wind-resources scene appearance is equal, is designated as p_kk=1/NN；According to { TT, WW }^befIn the corresponding global trend sequence of each wind-resources scene graph calculate general between the global trend sequence of wind-resources two-by-two Rate weight distance, is designated as PD_kk, formula is as follows：

${\mathrm{PD}}_{kk}=p_{kk}\&times;\underset{jj=1,jj\&NotEqual;kk}{\overset{KK}{\&Sigma;}}{D}_{kk-jj},$

WhereinKk ≠ jj, kk=1,2 ..., KK, jj=1,2 ..., KK；

For k scene of kth, the set containing KK-1 distance can be obtained, { D is designated as_kk-jj, by { D_kk-jjIn minimum value correspondence Scene sequence number jj be labeled as JL_kk, can be obtained containing KK JL for KK scene_kkSet, be designated as { JL_kk}；For epicycle Wind-resources scene set { TT, WW } before reduction^bef, the set containing KK probability right distance can be obtained, { PD is designated as_kk, will {PD_kkIn the corresponding scene sequence number kk of minimum value be labeled as PDmin.

5. the wind-resources scene reduction method according to claim 4 based on trend fitting, it is characterised in that step 6 is by such as Lower step is carried out：

By serial number PDmin scene from { TT, WW }^befIt is middle to delete；In { JL_kkIn find out PDmin element JL_PDmin, so Afterwards by the Probability p of serial number PDmin scene_PDminAdd to serial number JL_PDminScene probabilityI.e.So far, it can obtain the scene set { TT, WW } after epicycle reduction scene^aftAnd its corresponding probability； The number of times for performing this step is designated as mm, if mm<MM, then go to step 5 and start new round reduction, otherwise end step 5 is to step Rapid 6 circulation, is met scene set and its probability of scene reduction number requirement.