A distributed photovoltaic group control method

Abstract

A distributed photovoltaic group modulation and control method belongs to the technical field of group modulation and control. It improves the initial spacing by obtaining a time-space correlation improvement coefficient and an active coordination time-space criticality coefficient, obtains the time-space connection between multiple information through the time-space correlation improvement coefficient to identify the output voltage value of the outlier that is continuously maintained at a time point but changes slowly; configures its sensitivity according to some mobile attributes through the active coordination time-space criticality coefficient, so as to more accurately obtain the complex outlier output voltage value pattern; therefore, the present invention can efficiently handle the complexity and variability in the distributed photovoltaic cluster scenario, whether it is a distributed photovoltaic system with a small dust concentration in the surrounding air or a distributed photovoltaic system with a large dust concentration in the surrounding air, the method can maintain an efficient outlier monitoring function, improve the monitoring sensitivity to subtle output voltage changes, and enhance the accuracy of identifying hidden distributed photovoltaic systems with abnormal operation.

Description

Technical Field

[0001] The present invention belongs to the technical field of group regulation and group control, and specifically relates to a distributed photovoltaic group regulation and group control method. Background Art

[0002] Distributed photovoltaic refers to photovoltaic power generation facilities that are built near the user's site, with the characteristics of self-generation and self-use on the user side, surplus electricity being connected to the grid, and balanced and regulated in the distribution system.

[0003] In terms of distributed photovoltaic group control, the existing technical solution with patent publication number "CN118523483A" is currently generally used to achieve it. It includes real-time sampling of the output voltage of each distributed photovoltaic system in the distributed photovoltaic cluster to determine the corresponding deviation voltage to calculate the corresponding power adjustment amount.

[0004] In addition, to improve the control function of distributed photovoltaic group dispatching and group control, it is often necessary to use the output voltage of each distributed photovoltaic system sampled in real time to estimate the abnormal operation of the distributed photovoltaic system. The existing estimation of the abnormal operation of the distributed photovoltaic system mainly relies on the independent output voltage sampling combined with the outlier detection model analyzed by the CLARANSI algorithm. However, this model has some limitations in actual application: the output voltage increase is not fast during the initial abnormal operation of the distributed photovoltaic system, and it often only increases by one to two volts per hour. Such slight changes are often misjudged by existing methods as reasonable operating fluctuations. In addition, the sunlight intensity is uneven for different distributed photovoltaic systems. The existing CLARANSI algorithm cannot accurately reflect the outlier trend of the overall output voltage. As a result, the existing method often ignores the defects of hidden operating abnormalities, which is not conducive to the accuracy and reliability of early monitoring of distributed photovoltaic system operating abnormalities. Summary of the invention

[0005] In order to solve the defects in the prior art, the present invention proposes a distributed photovoltaic group modulation and group control method, which improves the starting distance by obtaining the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient, obtains the time-space connection between multiple information through the time-space correlation improvement coefficient to identify the output voltage value of the outlier that is continuously maintained at a point in time but changes slowly; configures its sensitivity according to some mobile attributes through the active coordination time-space criticality coefficient, so as to more accurately obtain the complex outlier output voltage value pattern; therefore, the present invention can efficiently handle the complexity and variability of distributed photovoltaic cluster scenarios, whether it is a distributed photovoltaic system with a small dust concentration in the surrounding air or a distributed photovoltaic system with a large dust concentration in the surrounding air, the method can maintain an efficient outlier monitoring function, improve the monitoring sensitivity to subtle output voltage changes, and enhance the accuracy of identifying hidden distributed photovoltaic systems with abnormal operation.

[0006] The present invention uses the following technical solutions.

[0007] A distributed photovoltaic group modulation and group control method, comprising:

[0008] In a distributed photovoltaic cluster, the output voltage of each distributed photovoltaic system is sampled in real time to identify the corresponding deviation voltage and calculate the corresponding power adjustment amount;

[0009] The distributed photovoltaic group modulation and group control method also includes:

[0010] Step 1: At each sampling time point, obtain the multivariate information of the distributed photovoltaic system at different locations in the distributed photovoltaic cluster, and form a multivariate information queue with all the multivariate information in the pre-defined time interval. Here, each multivariate information includes the output voltage value, sampling time point and location coordinate value;

[0011] Step 2: When performing grouping on the multiple information in the multiple information queue, respectively obtain the starting distance between each pair of multiple information in the multiple information queue, and for any starting distance, obtain the time-space correlation improvement coefficient of the starting distance according to the information difference between the pair of multiple information corresponding to the starting distance, and obtain the active coordination time-space criticality coefficient of the starting distance according to the information attribute of the pair of multiple information corresponding to the starting distance;

[0012] Step 3: Use the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient to improve the starting interval, obtain the improved interval, obtain the corresponding improved interval of each starting interval, and cut the multi-information in the multi-information queue into more than one group according to the overall improved interval;

[0013] Step 4: Based on the multivariate information in each group, obtain the concentrated outlier index of each group, and select the multivariate information that generates outliers in the pre-defined time interval based on the concentrated outlier index of each group.

[0014] Furthermore, in Step 1, corresponding voltage transmitters and GPS modules connected to the industrial computer are provided on the distributed photovoltaic systems at different locations in the distributed photovoltaic cluster. The voltage transmitter samples the output voltage value of the corresponding distributed photovoltaic system once per second and transmits it to the industrial computer. The synchronous GPS module samples the location coordinate value of the corresponding distributed photovoltaic system and transmits it to the industrial computer. The industrial computer receives the output voltage values, sampling time points and location coordinate values ​​transmitted by the sampling as a multivariate information and defines it as: , obtain the multivariate information queue formed by arranging all the multivariate information in 24 hours in the order of their sampling time points. Here, It is the first Multiple information, is the sequence code of the multiple information in the multiple information queue, It is the first The output voltage value of the multivariate information, It is the first The location coordinates of multiple information, It is the first The sampling time point of multiple information.

[0015] Furthermore, in Step 2, the starting distance between each pair of multivariate information is the L2 norm between the pair of multivariate information.

[0016] Furthermore, in Step 2, facing an arbitrary starting distance, a method for obtaining a time-space correlation improvement coefficient of the starting distance includes:

[0017] Obtain the sampling time point subtraction of a pair of multivariate information corresponding to the starting interval, obtain the square value of the sampling time point subtraction divided by the square value of the pre-defined time point span parameter as the ratio 1, perform negative correlation normalization on the ratio 1, and the obtained value is regarded as the negative correlation normalization amount 1;

[0018] Obtain the L2 norm of the location coordinate values ​​between a pair of multivariate information corresponding to the starting interval, obtain the square value of the L2 norm divided by the square value of the location coordinate span parameter defined in advance as the ratio 2, perform negative correlation normalization on the ratio 2, and obtain the value as the negative correlation normalization amount 2;

[0019] The amount obtained by multiplying the negative correlation standardized amount 1 and the negative correlation standardized amount 2 is used as the time-space correlation improvement coefficient of the starting interval.

[0020] Furthermore, in Step 2, the multivariate information is calculated and multiple information The equation for the improvement coefficient of the time-space correlation of the starting interval is:

[0021] ,

[0022] Here, It is multi-information and multiple information Improvement factor of time-space correlation of starting distance; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; , Multiple Information and multiple information The sampling time point; It is a predefined time span parameter; , Multiple Information and multiple information The location coordinates of It is the pre-defined location coordinate span parameter; is the Euler number.

[0023] Furthermore, in Step 2, the method for obtaining the criticality coefficient of the active coordination of the starting interval includes:

[0024] Step 2-1: For a random multi-information in a pair of multi-information corresponding to the starting interval, obtain the output voltage subtraction amount of the multi-information and all the multi-information in the multi-information queue other than the multi-information, form a neighboring group of the multi-information with the output voltage subtraction amount lower than the defined output voltage subtraction threshold, and obtain a partial grouping factor of the multi-information according to the neighboring group of the multi-information;

[0025] Step 2-2: Obtaining a partial time point continuity factor of the multi-dimensional information according to the output voltage value of the multi-dimensional information that is the same as the location coordinate value of the multi-dimensional information;

[0026] Step 2-3: Obtain partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval respectively, and obtain the active coordination time-space criticality coefficient of the starting interval based on the partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval.

[0027] Furthermore, in Step 2-1, the method for obtaining partial grouping factors of multivariate information includes:

[0028] Obtain the number of multivariate information in the neighboring group of multivariate information, calculate the subtraction between the number of multivariate information in the neighboring group of multivariate information and a constant 1, obtain the subtraction amount 1, obtain the multiplication of the subtraction amount 1 and the number of multivariate information in the neighboring group of multivariate information as the multiplication amount 1;

[0029] respectively obtaining the output voltage value subtraction amount between each pair of multi-element information in the neighboring group of multi-element information, and obtaining the number of output voltage value subtraction amounts that are all below the output voltage subtraction threshold amount in the neighboring group of multi-element information;

[0030] The ratio obtained by dividing the number of output voltage subtractions of a predefined increase rate by the multiplication amount 1 is obtained, and the obtained ratio value is used as a partial grouping factor of the multivariate information.

[0031] Further, in Step 2-1, the number of output voltage subtractions in the output voltage subtraction queue is obtained and defined as , get nearby groups The number of internal multivariate information and the operation of multivariate information The equation for the partial grouping factor is:

[0032] ,

[0033] Here, It is multi-information Some grouping factors of ; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; It is multi-information nearby groups; is the number of multivariate information in the neighboring group of multivariate information; is the number of output voltage subtractions in the output voltage subtraction queue.

[0034] Further, in Step 2-2, in the predefined time interval, the sampling time point of the multivariate information and the predefined short time duration before the sampling time point of the multivariate information are taken as the predefined short time duration, and the multivariate information with the same location coordinate value as the multivariate information in the predefined short time duration is formed into a short time sub-queue, and the output voltage standard deviation of the short time sub-queue is obtained according to the output voltage value of the multivariate information in the short time sub-queue;

[0035] In a predefined time interval, the sampling time point of the multivariate information and the predefined long time before the sampling time point of the multivariate information are taken as the predefined long time, and the multivariate information with the same location coordinate value as the multivariate information in the predefined long time interval is formed into a long time sub-queue, and the output voltage standard deviation of the long time sub-queue is obtained according to the output voltage value of the multivariate information in the long time sub-queue;

[0036] The ratio of the output voltage standard deviation of the short-time sub-queue divided by the output voltage standard deviation of the long-time sub-queue is obtained as ratio three, and the subtraction of constant 1 and ratio three is obtained as the partial time point continuity factor of the multivariate information.

[0037] Furthermore, in Step 2-2, the multivariate information is calculated The equation for the partial time point persistence factor is:

[0038] ,

[0039] Here, It is multi-information The persistence factor of some time points; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; is the standard deviation of the output voltage of the short-time sub-queue; is the standard deviation of the output voltage of the long-time sub-queue; It is the output voltage value of the multivariate information.

[0040] Further, in Step 2-3, the mean of some grouping factors of a pair of multivariate information corresponding to the starting interval is taken as mean 1, and the sum of constant 1 and mean 1 is calculated to obtain sum 1; the mean of some time point continuity factors of a pair of multivariate information corresponding to the starting interval is taken as mean 2, and the sum of constant 1 and mean 2 is calculated to obtain sum 2; the output voltage subtraction of a pair of multivariate information corresponding to the starting interval is obtained, and the square value of the output voltage subtraction is obtained and divided by the pre-defined output voltage span parameter The ratio obtained by calculating the square value of is taken as ratio three, negative correlation normalization is performed on ratio three, and the obtained value is taken as negative correlation normalized quantity three, the subtraction between constant 1 and negative correlation normalized quantity three is calculated to obtain subtraction quantity one, and the subtraction quantity one of the pre-defined ratio is obtained as the disordered subtraction quantity; the quantity obtained by multiplying the sum value one, the sum value two and the disordered subtraction quantity is calculated to obtain multiplication quantity two, and the quantity obtained by multiplying the pre-defined time and place critical coefficient parameter and multiplication quantity two is obtained as the active coordinated time and place critical coefficient of the starting interval.

[0041] Furthermore, in Step 2-3, the multivariate information is calculated and multiple information The equation for the criticality coefficient of the active coordination of the starting spacing is:

[0042] ,

[0043] Here, It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is a predefined time and place criticality coefficient parameter; , Multiple Information and multiple information Some grouping factors of ; , Multiple Information and multiple information The persistence factor of some time points; , Multiple Information and multiple information Output voltage value; is the predefined output voltage span parameter, is the Euler number.

[0044] Furthermore, in Step 3, the amount obtained by multiplying the correlation improvement coefficient during calculation and the criticality coefficient during active coordination is used as multiplication amount three, the subtraction amount between the constant 1 and the multiplication amount three is calculated to obtain the subtraction amount two, and the amount obtained by multiplying the starting distance and the subtraction amount two is used as the improvement distance.

[0045] Furthermore, in Step 3, the equation for calculating the improved spacing is:

[0046] ,

[0047] Here, It is multi-information and multiple information Improved spacing; It is multi-information and multiple information The starting distance of It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is multi-information and multiple information Improvement factor of time-space correlation of starting distance.

[0048] Furthermore, in Step 3, after obtaining the improved interval after improving the starting interval, the improved intervals corresponding to each starting interval are obtained, and the overall improved intervals are used as the intervals used by the CLARANSI algorithm. The CLARANSI algorithm is then executed on the multi-information queue to cut the multi-information in the multi-information queue into more than one group.

[0049] Furthermore, in Step 4, the attributes of each group are calculated, which include the mean output voltage value of each group and the number of multivariate information in the group, and the benchmark index of each group attribute is calculated. The centralized outlier index of each group is calculated through the benchmark index of each group attribute, and the multivariate information that generates outliers is selected. The distributed photovoltaic system corresponding to the multivariate information that generates outliers is a distributed photovoltaic system with hidden dangers of abnormal operation.

[0050] Furthermore, in Step 4, the method of calculating the grouped concentrated outlier index includes:

[0051] For any group, the output voltage reference index of the group is obtained according to the output voltage value of the multivariate information in the group, the number reference index of the group is obtained according to the number of the multivariate information in the group, and the exponential subtraction between the output voltage reference index of the group and the number reference index of the group is taken as the concentrated outlier index of the group.

[0052] Furthermore, in Step 4, the calculation equation of the base index of the grouping attribute is:

[0053] ,

[0054] Here , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces; It is The average output voltage value of each group; and are the mean and standard deviation of the output voltage values ​​of all groups respectively; It is The number of multivariate information in each group; and are the mean and standard deviation of the number of multivariate information in all groups.

[0055] Furthermore, in Step 4, after obtaining the benchmark index of the grouping attribute, the equation for calculating the concentrated outlier index of the grouping according to the benchmark index of the grouping attribute is:

[0056] ,

[0057] Here, It is The concentrated outlier index of the grouping; , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces.

[0058] Furthermore, in Step 4, the concentrated outlier index of each group is obtained, and each group is arranged from high to low according to its concentrated outlier index. Among the arranged groups, the group with a pre-defined ratio is selected as the outlier group, and the multivariate information in the outlier group is regarded as the multivariate information generated by the outlier in the pre-defined time interval.

[0059] A distributed photovoltaic group modulation and group control device, comprising:

[0060] A formation module is used to obtain multivariate information of distributed photovoltaic systems at different locations in the distributed photovoltaic cluster at each sampling time point, and form all multivariate information in a predefined time interval into a multivariate information queue, where each multivariate information includes an output voltage value, a sampling time point and a location coordinate value;

[0061] A grouping module, which is used to obtain the starting distance between each pair of multiple information in the multiple information queue when grouping multiple information in the multiple information queue, and obtain the time-space correlation improvement coefficient of the starting distance according to the information difference between the pair of multiple information corresponding to the starting distance for any starting distance, and obtain the active coordination time-space criticality coefficient of the starting distance according to the information attribute of the pair of multiple information corresponding to the starting distance;

[0062] An improvement module, which is used to improve the starting interval by using the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient, obtain the improvement interval, obtain the improvement interval corresponding to each starting interval, and cut the multi-information in the multi-information queue into more than one group according to the overall improvement interval;

[0063] The outlier module is used to obtain the concentrated outlier index of each group according to the multivariate information in each group, and select the multivariate information that generates outliers in a predefined time interval according to the concentrated outlier index of each group.

[0064] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0065] At each sampling time point, obtain the multivariate information of the distributed photovoltaic system at different locations in the distributed photovoltaic cluster, and form a multivariate information queue with all the multivariate information in the predefined time interval; when grouping the multivariate information in the multivariate information queue, respectively obtain the starting intervals between each pair of multivariate information in the multivariate information queue, and for any starting interval, obtain the time-space correlation improvement coefficient of the starting interval according to the information difference between the pair of multivariate information corresponding to the starting interval, and obtain the active coordination time-space criticality coefficient of the starting interval according to the information attributes of the pair of multivariate information corresponding to the starting interval; use the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient to improve the starting interval, obtain the improved interval, obtain the improved interval corresponding to each starting interval, and cut the multivariate information in the multivariate information queue into more than one group according to the total improved interval; obtain each multivariate information in each group according to the multivariate information The concentrated outlier index of the grouping selects the multivariate information that generates outliers in the pre-defined time interval. The present invention improves the starting interval by obtaining the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient, and obtains the time-space connection between the multivariate information through the time-space correlation improvement coefficient to identify the output voltage value of the outlier that is continuously maintained at a point in time but changes slowly; through the active coordination of the time-space criticality coefficient, the sensitivity is configured according to the partial mobility attributes, so as to more accurately obtain the complex output voltage value pattern of the outlier; therefore, the present invention can efficiently handle the complexity and variability of the distributed photovoltaic cluster scenario, whether it is a distributed photovoltaic system with a small dust concentration in the surrounding air or a distributed photovoltaic system with a large dust concentration in the surrounding air, the method can maintain an efficient outlier monitoring function, improve the monitoring sensitivity to subtle output voltage changes, and enhance the accuracy of identifying hidden distributed photovoltaic systems with abnormal operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a partial flow chart of the distributed photovoltaic group modulation and group control method described in the present invention;

[0067] Figure 2 It is a partial structural schematic diagram of the distributed photovoltaic group modulation and group control device described in the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely expressed in combination with the drawings in the embodiments of the present invention. The embodiments expressed in this application are only some embodiments of the present invention, not all embodiments. According to the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the protection scope of the present invention.

[0069] like Figure 1As shown, a distributed photovoltaic group modulation and group control method described in the present invention includes:

[0070] In a distributed photovoltaic cluster, the output voltage of each distributed photovoltaic system is sampled in real time to identify the corresponding deviation voltage and calculate the corresponding power adjustment amount;

[0071] The distributed photovoltaic group modulation and group control method also includes:

[0072] Step 1: At each sampling point in time, obtain the multivariate information of distributed photovoltaic systems at different locations in the distributed photovoltaic cluster, and form a multivariate information queue of all multivariate information in the pre-defined time interval. Here, each multivariate information includes output voltage value, sampling time point and location coordinate value; different distributed photovoltaic systems are in different locations, that is, distributed photovoltaic systems at different locations are different distributed photovoltaic systems.

[0073] In a preferred but non-limiting embodiment of the present invention, in Step 1, a voltage transmitter and a GPS module connected to a corresponding industrial computer (the distributed photovoltaic group adjustment and group control method is run on the industrial computer) are provided on distributed photovoltaic systems at different locations in the distributed photovoltaic cluster. The voltage transmitter samples the output voltage value of the corresponding distributed photovoltaic system once per second and transmits it to the industrial computer. The synchronous GPS module samples the location coordinate value of the corresponding distributed photovoltaic system and transmits it to the industrial computer. The industrial computer receives the output voltage values, sampling time points and location coordinate values ​​transmitted by the sampling as a multivariate information and defines it as: , obtain the multivariate information queue formed by arranging all the multivariate information in 24 hours in the order of their sampling time points. Here, It is the first Multiple information, It is the sequence code of the multiple information in the multiple information queue (the sequence code is set one by one according to the sampling time sequence of the multiple information), It is the first The output voltage value of the multivariate information, It is the first The location coordinates of multiple information, It is the first The sampling time point of multiple information.

[0074] Step 2: When performing grouping on the multiple information in the multiple information queue, respectively obtain the starting distance between each pair of multiple information in the multiple information queue, and for any starting distance, obtain the time-space correlation improvement coefficient of the starting distance according to the information difference between the pair of multiple information corresponding to the starting distance, and obtain the active coordination time-space criticality coefficient of the starting distance according to the information attribute of the pair of multiple information corresponding to the starting distance;

[0075] After obtaining the multivariate information queue, the CLARANSI algorithm is executed on the multivariate information in the multivariate information queue. During the grouping, the starting distance between each pair of multivariate information in the multivariate information queue can be obtained. In a preferred but non-restrictive embodiment of the present invention, in Step 2, the starting distance between each pair of multivariate information is the L2 norm between the pair of multivariate information. The existing estimation of the abnormal operation status of the distributed photovoltaic system is performed according to the starting distance to perform grouping analysis. According to the characteristics of the abnormal operation of the distributed photovoltaic system, the output voltage increases very slowly in the initial stage of the abnormal operation, and often only increases by one to two volts per hour. Such slight changes are often misjudged as reasonable operation fluctuations. The complicated installation positions of the synchronized distributed photovoltaic clusters and the dust concentration in the air make the sunlight intensity uneven for each distributed photovoltaic system in the distributed photovoltaic cluster, so that the outlier output voltage value represents a complicated and disordered arrangement, just like the distributed photovoltaic system at a location in a distributed photovoltaic cluster. The output voltage starts to increase slowly. Due to the difference in the installation positions of the surrounding distributed photovoltaic systems, the distributed photovoltaic systems around the location are facing different sunlight coverage areas, and the arrangement of their output voltages will show a disordered form. The output voltage of some distributed photovoltaic systems close to the location rises rapidly, while the output voltage of other distributed photovoltaic systems often remains basically unchanged. Such a complicated output voltage arrangement pattern makes it impossible to accurately obtain the outlier trend of the abnormal operation of the distributed photovoltaic system by the existing method of estimating the abnormal operation of the distributed photovoltaic system based on the output voltage value of the distributed photovoltaic system at a single location.

[0076] To overcome this defect, it is necessary to obtain the time-space correlation improvement coefficient of the starting interval and the active coordination time-space criticality coefficient, and obtain the time-space (time point and location) relationship between multiple information through the time-space correlation improvement coefficient to identify the output voltage value of the outlier that is maintained continuously at the time point but changes slowly; through the active coordination time-space criticality coefficient, the sensitivity is configured according to the local mobile attributes, so as to more accurately obtain the complex outlier output voltage value mode.

[0077] In the early stage monitoring of abnormal operation of distributed photovoltaic systems, it is necessary to accurately identify the outlier output voltage values ​​that are maintained continuously at a point in time but change slowly in the complex distributed photovoltaic cluster scenario. The key lies in the slowness of output voltage change, the unevenness of sunlight intensity, and the coherence of time and space. For a distributed photovoltaic cluster, the output voltage of a distributed photovoltaic system only increases by one volt per hour. Such slight changes are often misjudged as reasonable operating fluctuations or sampling errors. Existing monitoring methods often take dozens or even hundreds of hours to confirm whether such changes are traceable. Such delays often miss the best maintenance period for distributed photovoltaic systems. This is because the geographical locations of distributed photovoltaic clusters are complex and different distributed photovoltaic systems are installed in different locations, which makes the sunlight intensity uneven, just as in a location with abnormal operation. When the distributed photovoltaic system is installed facing the sunlight, the output voltage of the distributed photovoltaic system often increases at a higher rate. When the distributed photovoltaic system is installed away from the sunlight, the output voltage of the distributed photovoltaic system often increases at a very slow rate. This imbalance makes it impossible for the existing method that only relies on the output voltage value of the distributed photovoltaic system at a single location to accurately explain the overall outlier output voltage trend. In addition, the operation anomaly of the distributed photovoltaic system is a continuous process in both time and location. The output voltage change of a distributed photovoltaic system often affects the surrounding distributed photovoltaic systems, and this effect will take some time to become apparent. A slight output voltage increase is monitored on a distributed photovoltaic system in a distributed photovoltaic cluster. Therefore, as the time point moves backward, this outlier output voltage value will spread to the surrounding distributed photovoltaic systems.

[0078] To identify the outlier output voltage value that is maintained continuously at a time point but changes slowly, it is necessary to obtain the time-location correlation improvement coefficient of the starting interval. The time-location correlation improvement coefficient not only involves the output voltage value of a single multivariate information, but also concentrates the connection between the multivariate information and other multivariate information at the time point and the location. For example, the output voltage of the distributed photovoltaic system at a monitoring location has been one volt higher than the previous average in the previous six hours. Therefore, even if the difference is very low, it will be identified as something to be taken seriously. Through this method, a slow but continuous output voltage increase trend can be efficiently obtained; similarly, even if a distributed photovoltaic system at a single monitoring location does not show a significant outlier output voltage value, several monitoring locations around it have slight output voltage increases, which will also be identified as something to be taken seriously. Therefore, in a preferred but non-limiting embodiment of the present invention, in Step 2, facing an arbitrary starting interval, a method for obtaining the time-location correlation improvement coefficient of the starting interval includes:

[0079] Obtain a subtraction amount obtained by subtracting a pair of sampling time points of multivariate information corresponding to the starting interval, obtain a ratio obtained by dividing the square value of the sampling time point subtraction amount by the square value of the pre-defined time point span parameter as ratio one, perform negative correlation normalization processing on ratio one, and obtain a value as negative correlation normalization amount one;

[0080] Obtain the L2 norm of the location coordinate values ​​between a pair of multivariate information corresponding to the starting interval, obtain the square value of the L2 norm divided by the square value of the location coordinate span parameter defined in advance as the ratio 2, perform negative correlation normalization on the ratio 2, and obtain the value as the negative correlation normalization amount 2;

[0081] The amount obtained by multiplying the negative correlation standardized amount 1 and the negative correlation standardized amount 2 is used as the time-space correlation improvement coefficient of the starting interval.

[0082] In a preferred but non-limiting embodiment of the present invention, in Step 2, multiple information is used and multiple information For the starting distance, the multivariate information and multiple information The equation for the improvement coefficient of the time-space correlation of the starting interval is:

[0083] ,

[0084] Here, It is multi-information and multiple information Improvement factor of time-space correlation of starting distance; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; , Multiple Information and multiple information The sampling time point; It is a predefined time span parameter; , Multiple Information and multiple information The location coordinates of It is the pre-defined location coordinate span parameter; is the Euler number.

[0085] The predefined time span parameter controls the rate at which the time correlation decreases, along with the time point subtraction. Increase, time point correlation As it approaches zero, the improvement factor of the time-space correlation of the initial distance decreases; the predefined location coordinate span parameter controls the gradual deceleration rate of location correlation, along with the location coordinate spacing. Increased, location-related As it approaches zero, the improvement coefficient of the time-space correlation of the starting interval decreases; the pre-defined time point span parameter can be defined as one hour, and the pre-defined location coordinate span parameter can be defined as sixty meters.

[0086] For example, in a distributed photovoltaic cluster, even if the output voltage increases by only one volt per hour, by configuring the time span parameter It can make the time-location correlation improvement coefficient maintain a high value in a relatively short time span, that is, even if a pair of multivariate information is several hours apart, its time correlation is still very close, so under the condition of continuous increase in output voltage, even if the increase is very low, the time-location correlation improvement coefficient will also maintain a high value, so that such subtle but continuous changes are highlighted; in the face of the uneven intensity of sunlight, when the installation position of the surrounding distributed photovoltaic system is facing the sunlight direction, the output voltage of the distributed photovoltaic system often increases at a higher rate, by configuring the location span parameter The multivariate information with different location coordinates still has a close correlation, and the time-place correlation improvement coefficient will also maintain a high value. The multivariate information whose installation position is facing the sunlight direction will be grouped into a group, which shows that the output voltage of the distributed photovoltaic system often increases at a higher rate; if a pair of multivariate information is close in time and location, its time-place correlation improvement coefficient will be high, which means that in the sampled multivariate information queue, if the output voltage of a distributed photovoltaic system starts to increase and is accompanied by a delay in time, the correlation of the multivariate information of the surrounding distributed photovoltaic systems will be large, reflecting the expansion mode of the outlier output voltage value; in the face of short-term, local output voltage fluctuations (such as temporary voltage boost caused by the operation of the photovoltaic control system), even if the location correlation is often large, the time correlation will be small. Under such conditions, the value of the time-place correlation improvement coefficient cannot be too high to avoid the defect of misidentifying short-term fluctuations as hidden danger information.

[0087] There are also the following defects in the scenario of abnormal operation monitoring of distributed photovoltaic systems:

[0088] Because the output voltage distribution in the distributed photovoltaic cluster scenario is affected by several factors and the factors have different manifestations in different distributed photovoltaic systems and periods, just as the output voltage of the distributed photovoltaic system with a smaller dust concentration in the surrounding air in the distributed photovoltaic cluster is relatively stable, the output voltage of the distributed photovoltaic system with a larger dust concentration in the surrounding air will fluctuate more. Distributed photovoltaic systems in different locations often show different output voltage change patterns due to the effects of surrounding humidity and air pressure. In addition, the operation of components in the distributed photovoltaic cluster will also have a significant effect on the output voltage distribution. When the load change frequency faced by the components in the distributed photovoltaic cluster is not small, the output voltage will show more complex disordered changes. Therefore, if the same outlier output voltage value identification principle is applied to all distributed photovoltaic systems, a lot of misidentification will occur in some distributed photovoltaic systems, and important abnormal information will often be missed in other distributed photovoltaic systems. For example, the output voltage increase of 5 volts in a distributed photovoltaic system with a small dust concentration in the surrounding air in the distributed photovoltaic cluster will not reach the outlier critical value, but the output voltage increase of the same amplitude in a distributed photovoltaic system with a large dust concentration in the surrounding air will cause reliability risks.

[0089] To overcome the above defects, it is necessary to obtain the active coordinated time-space criticality coefficient of the starting interval, and configure its sensitivity according to the local mobility attributes through the active coordinated time-space criticality coefficient, so as to more accurately obtain the complex outlier output voltage value pattern. To obtain the active coordinated time-space criticality coefficient, a pair of important factors are involved: local grouping attributes and time point continuity. The local grouping attributes reflect the dense amplitude of the output voltage value in the local interval. Distributed photovoltaic systems around hidden abnormal operation locations often have output voltage values ​​that represent a certain amount of density, just like a distributed photovoltaic system in a distributed photovoltaic cluster. If there are one hundred square meters around the distributed photovoltaic system, The distributed photovoltaic systems at several monitoring locations in the distributed photovoltaic system all show slight but similar output voltage increases. Therefore, even if the amplitude of such an increase is not obvious in the distributed photovoltaic cluster, it is also a piece of news that can be paid attention to. The time point continuity is obtained based on the continuity of the output voltage change. The operation abnormality of the distributed photovoltaic system is often a slow but continuous stage. It is necessary to pay special attention to the outlier output voltage value that is small but sustained. That is, if the output voltage of the distributed photovoltaic system at a monitoring location has been two volts higher than its long-term average in the past twelve hours, even if the difference looks small, it often indicates a hidden defect.

[0090] By actively coordinating the above factors, the criticality coefficient can automatically configure its identification principle according to the attributes of the current value. It will improve sensitivity when monitoring a very dense and partially dense distributed photovoltaic system, and will also obtain subtle but hidden potential output voltage changes. In a distributed photovoltaic system with a very balanced output voltage distribution, it will weaken sensitivity and reduce the probability of misdetermination. In a specific scenario, when a distributed photovoltaic cluster starts to connect to a new load or the dust concentration in the surrounding air changes, the output voltage distribution pattern often changes significantly. The existing constant parameter method requires manual configuration to be suitable for such changes. The active coordination of the criticality coefficient can automatically configure its parameters to maintain efficient outlier monitoring function, and can more accurately identify various types of outlier output voltage values. Whether it is a slow but continuous output voltage increase or a partial sudden output voltage increase, it can be appropriately grouped, improving the early identification function of hidden distributed photovoltaic system operation abnormal risks.

[0091] In a preferred but non-limiting embodiment of the present invention, in Step 2, the method for obtaining the criticality coefficient of the active coordination of the starting interval includes:

[0092] Step 2-1: For any multi-information in a pair of multi-information corresponding to the starting interval, obtain a subtraction amount obtained by subtracting the output voltages of the multi-information and all the multi-information in the multi-information queue other than the multi-information, form a neighboring group of multi-information with the output voltage subtraction amount lower than a defined output voltage subtraction threshold, and obtain a partial grouping factor of the multi-information according to the neighboring group of the multi-information;

[0093] In a preferred but non-limiting embodiment of the present invention, in Step 2-1, the method for obtaining partial grouping factors of multivariate information includes:

[0094] The number of multivariate information in the neighboring group of multivariate information is obtained, a subtraction amount is calculated by subtracting the number of multivariate information in the neighboring group of multivariate information from a constant 1, and a subtraction amount of 1 is obtained, and a multiplication amount of the subtraction amount of 1 and the number of multivariate information in the neighboring group of multivariate information is obtained as a multiplication amount of 1;

[0095] Obtaining a subtraction amount obtained by subtracting the output voltage values ​​of each pair of multi-element information in a nearby group of multi-element information, and obtaining the number of subtraction amounts obtained by subtracting the output voltage values ​​of all the multi-element information in the nearby group that are lower than the output voltage subtraction threshold;

[0096] The ratio obtained by dividing the number of output voltage subtractions of a predefined increase rate by the multiplication amount 1 is obtained, and the obtained ratio value is used as a partial grouping factor of the multivariate information.

[0097] Using multiple information and multiple information For the starting distance, facing multiple information , obtain multiple information and multiple information queues with multiple information internal The output voltage subtraction amount of all multivariate information outside the output voltage subtraction amount is lower than the defined output voltage subtraction threshold amount. The multi-information forms a multi-information Nearby Groups , where the output voltage is subtracted by the critical amount It can be defined as five volts, and the output voltage difference between each pair of multiple information in the nearby group is obtained, and the nearby group The output voltage subtraction amounts that are all lower than the output voltage subtraction threshold amount form an output voltage subtraction amount queue (the method of forming the output voltage subtraction amount queue can be to arrange the output voltage subtraction amounts according to the sampling order of the output voltage values ​​used as subtrahends). In a preferred but non-limiting embodiment of the present invention, in Step 2-1, the number of output voltage subtraction amounts in the output voltage subtraction amount queue is obtained and defined as , get nearby groups The number of internal multivariate information and the operation of multivariate information The equation for the partial grouping factor is:

[0098] ,

[0099] Here, It is multi-information Some grouping factors of ; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; It is multi-information nearby groups; is the number of multivariate information in the neighboring group of multivariate information; is the number of output voltage subtractions in the output voltage subtraction queue (i.e., the multivariate information Nearby Groups The number of output voltage subtractions within the output voltage subtraction threshold).

[0100] The partial grouping factor reflects the intensive amplitude of multivariate information in a partial interval. When the number of multivariate information in a nearby group is constant, the larger the output voltage subtraction amount below the output voltage subtraction critical amount in the nearby group (that is, the larger the number of output voltage subtraction amounts in the output voltage subtraction amount queue), the higher the partial grouping factor, and the larger the intensive amplitude of the multivariate information in the interval of the nearby group; the lower the output voltage subtraction amount below the output voltage subtraction critical amount in the nearby group (that is, the lower the number of output voltage subtraction amounts in the output voltage subtraction amount queue), the lower the partial grouping factor, and the smaller the intensive amplitude of the multivariate information in the interval of the nearby group.

[0101] Step 2-2: Obtaining a partial time point continuity factor of the multi-dimensional information according to the output voltage value of the multi-dimensional information that is the same as the location coordinate value of the multi-dimensional information;

[0102] In a preferred but non-limiting embodiment of the present invention, in Step 2-2, in a predefined time interval, the sampling time point where the multivariate information is located and the predefined short time duration before the sampling time point where the multivariate information is located are obtained as the predefined short time duration, and the multivariate information with the same location coordinate value as the multivariate information in the predefined short time duration is formed into a short time sub-queue, and the output voltage standard deviation of the short time sub-queue is obtained according to the output voltage value of the multivariate information in the short time sub-queue;

[0103] In a predefined time interval, the sampling time point of the multivariate information and the predefined long time before the sampling time point of the multivariate information are taken as the predefined long time, and the multivariate information with the same location coordinate value as the multivariate information in the predefined long time interval is formed into a long time sub-queue, and the output voltage standard deviation of the long time sub-queue is obtained according to the output voltage value of the multivariate information in the long time sub-queue;

[0104] The ratio obtained by dividing the output voltage standard deviation of the short-time sub-queue by the output voltage standard deviation of the long-time sub-queue is taken as ratio three, and the subtraction obtained by subtracting the constant 1 and ratio three is taken as the partial time point continuity factor of the multivariate information.

[0105] Using multiple information For example, multiple information The sampling time point is , in a predefined time interval, obtain and Treat the previous hour as a short-term predefined duration, and neutralize the short-term predefined duration with multiple information The multivariate information with the same location coordinate value forms a short-time sub-queue, and the output voltage standard deviation of the multivariate information in the short-time sub-queue is calculated according to the output voltage value of the multivariate information in the short-time sub-queue; and Treat the previous eight hours as a long pre-defined time, and neutralize the long pre-defined time with multiple information The multivariate information with the same location coordinate value forms a long-term sub-queue, and the output voltage standard deviation of the multivariate information in the long-term sub-queue is calculated based on the output voltage value of the multivariate information in the long-term sub-queue; here, one hour is a pre-defined short time duration, and eight hours is a pre-defined long time duration.

[0106] In a preferred but non-limiting embodiment of the present invention, in Step 2-2, the multivariate information is calculated The equation for the partial time point persistence factor is:

[0107] ,

[0108] Here, It is multi-information The persistence factor of some time points; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; is the standard deviation of the output voltage of the short-time sub-queue; is the standard deviation of the output voltage of the long-time sub-queue; It is the output voltage value of the multivariate information.

[0109] The multiple information Partial point persistence factor It is the ratio of the difference between the long-term standard deviation and the short-term standard deviation divided by the long-term standard deviation. Under the condition that the output voltage increases slowly but continuously in the early stage of abnormal operation, the long-term standard deviation will gradually increase, and the short-term standard deviation will be lower. The value of will be close to one, indicating a continuous change trend; under the condition that the output voltage remains stable for a long time, the output voltage standard deviation of the short-term sub-queue Approximate standard deviation of output voltage for long-time subqueues , The value of will be close to zero, indicating that there is no significant sustained trend change.

[0110] Step 2-3: Obtain partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval respectively, and obtain the active coordination time-space criticality coefficient of the starting interval based on the partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval.

[0111] In a preferred but non-limiting embodiment of the present invention, in Step 2-3, according to the partial grouping factor calculation equation and the partial time point continuity factor calculation equation, the partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval are obtained respectively; the mean of the partial grouping factors of a pair of multivariate information corresponding to the starting interval is obtained as mean one, and the sum of constant 1 and mean one is calculated to obtain sum one; the mean of the partial time point continuity factors of a pair of multivariate information corresponding to the starting interval is obtained as mean two, and the sum of constant 1 and mean two is calculated to obtain sum two; the output voltage phase of a pair of multivariate information corresponding to the starting interval is obtained. The subtraction amount obtained by subtracting the constant 1 from the negative correlation normalization amount 3 is obtained, and the ratio obtained by dividing the square value of the output voltage subtraction amount by the square value of the pre-defined output voltage span parameter is taken as ratio three. The negative correlation normalization treatment is performed on the ratio three, and the value obtained is taken as negative correlation normalization amount three. The subtraction amount obtained by subtracting the constant 1 from the negative correlation normalization amount 3 is obtained to obtain the subtraction amount 1, and the subtraction amount 1 of the pre-defined ratio is obtained as the disordered subtraction amount; the amount obtained by multiplying the sum value 1, the sum value 2 and the disordered subtraction amount is obtained to obtain the multiplication amount 2, and the amount obtained by multiplying the pre-defined time and place critical coefficient parameter and the multiplication amount 2 is taken as the active coordination time and place critical coefficient of the starting interval.

[0112] In a preferred but non-limiting embodiment of the present invention, in Step 2-3, multiple information is used and multiple information For the starting distance, the multivariate information and multiple information The equation for the criticality coefficient of the active coordination of the starting spacing is:

[0113] ,

[0114] Here, It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is a predefined time and place criticality coefficient parameter; , Multiple Information and multiple information Some grouping factors of ; , Multiple Information and multiple information The persistence factor of some time points; , Multiple Information and multiple information Output voltage value; is the predefined output voltage span parameter, is the Euler number.

[0115] The unordered items Make the norm operation more sensitive to the output voltage difference. When the output voltage difference is low, the value of the disorder term is close to zero. As the output voltage difference increases, the value of the disorder term gradually approaches one. The pre-defined output voltage span parameter Can be defined as one volt, pre-defined time and place critical coefficient parameters It can be defined as one-half; the partial grouping factor reflects the intensive amplitude of the multivariate information in the partial interval. The higher the intensive amplitude of the multivariate information in the partial interval, the higher the criticality coefficient during active coordination; the partial time point continuity factor reflects the continuity of the output voltage change. The higher the continuity of the output voltage change in the partial interval, the higher the criticality coefficient during active coordination.

[0116] Therefore, the time-space correlation improvement coefficient of the starting interval and the time-space criticality coefficient of active coordination can be obtained.

[0117] Step 3: Use the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient to improve the starting interval, obtain the improved interval, obtain the corresponding improved interval of each starting interval, and cut the multi-information in the multi-information queue into more than one group according to the overall improved interval;

[0118] In a preferred but non-limiting embodiment of the present invention, in Step 3, after obtaining the time-space correlation improvement coefficient of the starting interval and the active coordination time-space criticality coefficient, the time-space correlation improvement coefficient of the starting interval and the active coordination time-space criticality coefficient are used to improve the starting interval to obtain an improved interval.

[0119] The amount obtained by multiplying the correlation improvement coefficient during calculation and the criticality coefficient during active coordination is used as the multiplication amount three, the subtraction amount obtained by subtracting the calculation constant 1 from the multiplication amount three is used to obtain the subtraction amount two, and the amount obtained by multiplying the initial distance and the subtraction amount two is used as the improvement distance.

[0120] In a preferred but non-limiting embodiment of the present invention, in Step 3, multiple information is used and multiple information For the starting spacing, the equation for calculating the improved spacing is:

[0121] ,

[0122] Here, It is multi-information and multiple information Improved spacing; It is multi-information and multiple information The starting distance of It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is multi-information and multiple information Improvement factor of time-space correlation of starting distance.

[0123] The final improvement interval involves the time and location correlation of multiple information by using the time-location correlation improvement coefficient. By using the active coordination time-location criticality coefficient, the sensitivity is flexibly configured according to the attributes of the current value. In monitoring the densely distributed photovoltaic system (that is, The output voltage of the distributed photovoltaic system is characterized by a large value, or the output voltage is characterized by a continuous outlier (that is, When the value is large, the criticality coefficient of active coordination will increase accordingly, making it more likely to discover hidden dangers in such distributed photovoltaic systems.

[0124] In a preferred but non-limiting embodiment of the present invention, in Step 3, after obtaining the improved spacing after the initial spacing is improved, the improved spacing corresponding to each initial spacing is obtained, and the overall improved spacing is used as the spacing used by the CLARANSI algorithm, and then the CLARANSI algorithm is executed on the multi-information queue to cut the multi-information in the multi-information queue into more than one group. Here, the selection of the number of groups is determined based on the silhouette coefficient method.

[0125] Thus, grouping according to the improved spacing can be obtained.

[0126] Step 4: Based on the multivariate information in each group, obtain the concentrated outlier index of each group, and select the multivariate information that generates outliers in the pre-defined time interval based on the concentrated outlier index of each group.

[0127] The CLARANSI algorithm executed by improving the spacing has enhanced the classification effect of the hidden danger values ​​with abnormal operation compared with the existing grouping algorithm. After specific application and restraint, the grouping value with outlier attributes will have a lower number and a larger output voltage value than the general grouping, and the hidden abnormal operation hidden dangers in the specific distributed photovoltaic cluster are often not large. Therefore, in a preferred but non-restrictive embodiment of the present invention, in Step 4, the attributes of each group are calculated, which contain the average output voltage value of each group and the number of multivariate information in the group, and the benchmark index of each grouping attribute is calculated. The centralized outlier index of each group is calculated through the benchmark index of each grouping attribute, and the misalignment amplitude of the grouping and the overall arrangement is estimated to achieve outlier monitoring facing the grouping value, and the multivariate information that generates the outlier is selected. The distributed photovoltaic system corresponding to the multivariate information that generates the outlier is a distributed photovoltaic system with abnormal operation hazards. Based on this, the estimated distributed photovoltaic system with abnormal operation hazards can be used to perform maintenance or repair in advance, so as to achieve monitoring of the hidden abnormal operation hazards in the distributed photovoltaic cluster.

[0128] In a preferred but non-limiting embodiment of the present invention, in Step 4, the method for calculating the centralized outlier index of the grouping includes:

[0129] For any random group, the output voltage reference index of the group is obtained according to the output voltage value of the multivariate information in the group, the number reference index of the group is obtained according to the number of the multivariate information in the group, and the exponential subtraction obtained by subtracting the output voltage reference index of the group from the number reference index of the group is taken as the concentrated outlier index of the group.

[0130] In a preferred but non-limiting embodiment of the present invention, in Step 4, For each group, the calculation equation of the benchmark index of the group attribute is:

[0131] ,

[0132] Here , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces; It is The average output voltage value of each group; and are the mean and standard deviation of the output voltage values ​​of all groups respectively; It is The number of multivariate information in each group; and are the mean and standard deviation of the number of multivariate information in all groups.

[0133] The higher the average output voltage value of the group is, the higher the corresponding output voltage reference index is; the greater the amount of multivariate information in the group is, the higher the corresponding number reference index is.

[0134] In a preferred but non-limiting embodiment of the present invention, in Step 4, after obtaining the base index of the grouping attribute, the equation for calculating the concentrated outlier index of the grouping according to the base index of the grouping attribute is:

[0135] ,

[0136] Here, It is The concentrated outlier index of the grouping; , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces.

[0137] The higher the mean output voltage value of the group, the higher the corresponding output voltage reference index and the larger the concentrated outlier index; the larger the amount of multivariate information in the group, the higher the corresponding number reference index and the smaller the concentrated outlier index.

[0138] In a preferred but non-limiting embodiment of the present invention, in Step 4, the concentrated outlier index of each group is obtained, and each group is arranged from high to low according to the level of its concentrated outlier index. The group with a predefined ratio is selected from the arranged groups as the outlier group, and the multivariate information in the outlier group is used as the multivariate information generated in the predefined time interval. Here, the predefined ratio can be defined as one twentieth, just as the number of groups after arrangement is forty, so one twentieth of the number of groups is selected using the predefined ratio, that is, two groups are selected in sequence according to the order of the arrangement of each group from high to low, that is, two groups corresponding to the two highest concentrated outlier indexes are selected.

[0139] The interval in this application is the distance, so the multivariate information of outliers generated in the pre-defined time interval is obtained through the grouped centralized outlier index.

[0140] like Figure 2 As shown, a distributed photovoltaic group modulation and group control device described in the present invention includes:

[0141] A formation module is used to obtain multivariate information of distributed photovoltaic systems at different locations in the distributed photovoltaic cluster at each sampling time point, and form all multivariate information in a predefined time interval into a multivariate information queue, where each multivariate information includes an output voltage value, a sampling time point and a location coordinate value;

[0142] A grouping module, which is used to obtain the starting distance between each pair of multiple information in the multiple information queue when grouping multiple information in the multiple information queue, and obtain the time-space correlation improvement coefficient of the starting distance according to the information difference between the pair of multiple information corresponding to the starting distance for any starting distance, and obtain the active coordination time-space criticality coefficient of the starting distance according to the information attribute of the pair of multiple information corresponding to the starting distance;

[0143] An improvement module, which is used to improve the starting interval by using the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient, obtain the improvement interval, obtain the improvement interval corresponding to each starting interval, and cut the multi-information in the multi-information queue into more than one group according to the overall improvement interval;

[0144] The outlier module is used to obtain the concentrated outlier index of each group according to the multivariate information in each group, and select the multivariate information that generates outliers in a predefined time interval according to the concentrated outlier index of each group.

[0145] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0146] At each sampling time point, obtain the multivariate information of the distributed photovoltaic system at different locations in the distributed photovoltaic cluster, and form a multivariate information queue with all the multivariate information in the predefined time interval; when grouping the multivariate information in the multivariate information queue, respectively obtain the starting intervals between each pair of multivariate information in the multivariate information queue, and for any starting interval, obtain the time-space correlation improvement coefficient of the starting interval according to the information difference between the pair of multivariate information corresponding to the starting interval, and obtain the active coordination time-space criticality coefficient of the starting interval according to the information attributes of the pair of multivariate information corresponding to the starting interval; use the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient to improve the starting interval, obtain the improved interval, obtain the improved interval corresponding to each starting interval, and cut the multivariate information in the multivariate information queue into more than one group according to the total improved interval; obtain each multivariate information in each group according to the multivariate information The concentrated outlier index of the grouping selects the multivariate information that generates outliers in the pre-defined time interval. The present invention improves the starting interval by obtaining the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient, and obtains the time-space connection between the multivariate information through the time-space correlation improvement coefficient to identify the output voltage value of the outlier that is continuously maintained at a point in time but changes slowly; through the active coordination of the time-space criticality coefficient, the sensitivity is configured according to the partial mobility attributes, so as to more accurately obtain the complex output voltage value pattern of the outlier; therefore, the present invention can efficiently handle the complexity and variability of the distributed photovoltaic cluster scenario, whether it is a distributed photovoltaic system with a small dust concentration in the surrounding air or a distributed photovoltaic system with a large dust concentration in the surrounding air, the method can maintain an efficient outlier monitoring function, improve the monitoring sensitivity to subtle output voltage changes, and enhance the accuracy of identifying hidden distributed photovoltaic systems with abnormal operation.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Even though the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not deviate from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A distributed photovoltaic group control method, characterized in that: include: In a distributed photovoltaic cluster, the output voltage of each distributed photovoltaic system is sampled in real time to identify the corresponding deviation voltage and calculate the corresponding power adjustment amount; The distributed photovoltaic group modulation and group control method also includes: Step 1: At each sampling time point, obtain the multivariate information of the distributed photovoltaic system at different locations in the distributed photovoltaic cluster, and form a multivariate information queue with all the multivariate information in the pre-defined time interval. Here, each multivariate information includes the output voltage value, sampling time point and location coordinate value; Step 2: When performing grouping on the multiple information in the multiple information queue, respectively obtain the starting distance between each pair of multiple information in the multiple information queue, and for any starting distance, obtain the time-space correlation improvement coefficient of the starting distance according to the information difference between the pair of multiple information corresponding to the starting distance, and obtain the active coordination time-space criticality coefficient of the starting distance according to the information attribute of the pair of multiple information corresponding to the starting distance; Step 3: Use the time-space correlation improvement coefficient and the active coordination time-space criticality coefficient to improve the starting interval, obtain the improved interval, obtain the corresponding improved interval of each starting interval, and cut the multi-information in the multi-information queue into more than one group according to the overall improved interval; Step 4: Based on the multivariate information in each group, obtain the concentrated outlier index of each group, and select the multivariate information that generates outliers in the pre-defined time interval according to the concentrated outlier index of each group; In Step 2, for any starting distance, the method for obtaining the time-space correlation improvement coefficient of the starting distance includes: Obtain the sampling time point subtraction of a pair of multivariate information corresponding to the starting interval, obtain the square value of the sampling time point subtraction divided by the square value of the pre-defined time point span parameter as the ratio 1, perform negative correlation normalization on the ratio 1, and the obtained value is regarded as the negative correlation normalization amount 1; Obtain the L2 norm of the location coordinate values ​​between a pair of multivariate information corresponding to the starting interval, obtain the square value of the L2 norm divided by the square value of the location coordinate span parameter defined in advance as the ratio 2, perform negative correlation normalization on the ratio 2, and obtain the value as the negative correlation normalization amount 2; The amount obtained by multiplying the negative correlation standardized amount 1 and the negative correlation standardized amount 2 is taken as the time-space correlation improvement coefficient of the initial interval; In Step 2, calculate the multivariate information and multiple information The equation for the improvement coefficient of the time-space correlation of the starting interval is: Here, It is multiple information and multiple information Improvement factor of time-space correlation of starting distance; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; , Multiple Information and multiple information The sampling time point; It is a predefined time span parameter; , Multiple Information and multiple information The location coordinates of It is the pre-defined location coordinate span parameter; is the Euler number.

2. The distributed photovoltaic group control method according to claim 1 is characterized in that: In Step 1, corresponding voltage transmitters and GPS modules connected to the industrial computer are provided on the distributed photovoltaic systems at different locations in the distributed photovoltaic cluster. The voltage transmitter samples the output voltage value of the corresponding distributed photovoltaic system once per second and transmits it to the industrial computer. The synchronous GPS module samples the location coordinate value of the corresponding distributed photovoltaic system and transmits it to the industrial computer. The industrial computer receives the output voltage values, sampling time points and location coordinate values ​​transmitted by the sampling as a multivariate information and defines it as: , obtain the multivariate information queue formed by arranging all the multivariate information in 24 hours in the order of their sampling time points. Here, It is the first Multiple information, is the sequence code of the multiple information in the multiple information queue, It is the first The output voltage value of the multivariate information, It is the first The location coordinates of multiple information, It is the first The sampling time point of multiple information.

3. The distributed photovoltaic group control method according to claim 2 is characterized in that: In Step 2, the starting distance between each pair of multivariate information is the L2 norm between the pair of multivariate information.

4. The distributed photovoltaic group control method according to claim 3 is characterized in that: In Step 2, the method for obtaining the criticality coefficient of the active coordination of the starting interval includes: Step 2-1: For a random multi-information in a pair of multi-information corresponding to the starting interval, obtain the output voltage subtraction amount of the multi-information and all the multi-information in the multi-information queue other than the multi-information, form a neighboring group of the multi-information with the output voltage subtraction amount lower than the defined output voltage subtraction threshold, and obtain a partial grouping factor of the multi-information according to the neighboring group of the multi-information; Step 2-2: Obtaining a partial time point continuity factor of the multi-dimensional information according to the output voltage value of the multi-dimensional information that is the same as the location coordinate value of the multi-dimensional information; Step 2-3: Obtain partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval respectively, and obtain the active coordination time-space criticality coefficient of the starting interval based on the partial grouping factors and partial time point continuity factors of a pair of multivariate information corresponding to the starting interval.

5. The distributed photovoltaic group control method according to claim 4 is characterized in that: In Step 2-1, the method of obtaining partial grouping factors of multivariate information includes: Obtain the number of multivariate information in the neighboring group of multivariate information, calculate the subtraction between the number of multivariate information in the neighboring group of multivariate information and a constant 1, obtain the subtraction amount 1, obtain the multiplication of the subtraction amount 1 and the number of multivariate information in the neighboring group of multivariate information as the multiplication amount 1; respectively obtaining the output voltage value subtraction amount between each pair of multi-element information in the neighboring group of multi-element information, and obtaining the number of output voltage value subtraction amounts that are all below the output voltage subtraction threshold amount in the neighboring group of multi-element information; Obtaining a ratio obtained by dividing the number of output voltage subtractions of a predefined increase rate by the multiplication amount 1, and using the obtained ratio value as a partial grouping factor of the multivariate information; In Step 2-1, the number of output voltage subtractions in the output voltage subtraction queue is obtained and defined as , get nearby groups The number of internal multivariate information and the operation of multivariate information The equation for the partial grouping factor is: Here, It is multiple information Some grouping factors of ; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; It is multiple information nearby groups; is the number of multivariate information in the neighboring group of multivariate information; is the number of output voltage subtractions in the output voltage subtraction queue; In Step 2-2, in the predefined time interval, the sampling time point of the multivariate information and the predefined short time duration before the sampling time point of the multivariate information are taken as the predefined short time duration, and the multivariate information with the same location coordinate value as the multivariate information in the predefined short time duration is formed into a short time sub-queue, and the output voltage standard deviation of the short time sub-queue is obtained according to the output voltage value of the multivariate information in the short time sub-queue; In a predefined time interval, the sampling time point of the multivariate information and the predefined long time before the sampling time point of the multivariate information are taken as the predefined long time, and the multivariate information with the same location coordinate value as the multivariate information in the predefined long time interval is formed into a long time sub-queue, and the output voltage standard deviation of the long time sub-queue is obtained according to the output voltage value of the multivariate information in the long time sub-queue; The ratio obtained by dividing the output voltage standard deviation of the short-time sub-queue by the output voltage standard deviation of the long-time sub-queue is taken as ratio 3, and the subtraction between constant 1 and ratio 3 is taken as the partial time point persistence factor of the multivariate information; In Step 2-2, calculate the multivariate information The equation for the partial time point persistence factor is: Here, It is multi-information The persistence factor of some time points; It is the first Multiple information; It is the sequence code of the multiple information in the multiple information queue; is the standard deviation of the output voltage of the short-time sub-queue; is the standard deviation of the output voltage of the long-time sub-queue; is the output voltage value of the multivariate information; In Step 2-3, the mean of some grouping factors of a pair of multivariate information corresponding to the starting interval is obtained as mean 1, and the sum of constant 1 and mean 1 is calculated to obtain sum 1; the mean of some time point continuity factors of a pair of multivariate information corresponding to the starting interval is obtained as mean 2, and the sum of constant 1 and mean 2 is calculated to obtain sum 2; the output voltage subtraction of a pair of multivariate information corresponding to the starting interval is obtained, and the ratio obtained by dividing the square value of the output voltage subtraction by the square value of the pre-defined output voltage span parameter is obtained as ratio 3, negative correlation normalization is performed on ratio 3, and the obtained value is used as negative correlation normalization amount 3, and the subtraction of constant 1 and negative correlation normalization amount 3 is calculated to obtain subtraction amount 1, and the subtraction amount 1 of the pre-defined ratio is obtained as disordered subtraction; the amount obtained by multiplying the sum 1, sum 2 and disordered subtraction is obtained to obtain multiplication amount 2, and the amount obtained by multiplying the pre-defined time and place key coefficient parameter and multiplication amount 2 is obtained as the active coordination time and place key coefficient of the starting interval; In Step 2-3, calculate the multivariate information and multiple information The equation for the criticality coefficient of the active coordination of the starting spacing is: Here, It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is a predefined time and place criticality coefficient parameter; , Multiple Information and multiple information Partial grouping factors of ; , Multiple Information and multiple information The persistence factor of some time points; , Multiple Information and multiple information Output voltage value; is the predefined output voltage span parameter, is the Euler number.

6. The distributed photovoltaic group control method according to claim 5 is characterized in that: In Step 3, the amount obtained by multiplying the correlation improvement coefficient at the time of calculation and the key coefficient at the time of active coordination is used as the multiplication amount 3, the subtraction amount between the constant 1 and the multiplication amount 3 is calculated, and the subtraction amount 2 is obtained, and the amount obtained by multiplying the initial distance and the subtraction amount 2 is used as the improvement distance; In Step 3, the equation for calculating the improved spacing is: Here, It is multi-information and multiple information Improved spacing; It is multi-information and multiple information The starting distance of It is the first Multiple information; It is the first Multiple information; , It is the sequence code of the multiple information in the multiple information queue; It is multi-information and multiple information The criticality coefficient of the active coordination of the starting spacing; It is multi-information and multiple information Improvement factor of time-space correlation of starting distance; In Step 3, after obtaining the improved interval after the initial interval is improved, the improved intervals corresponding to each initial interval are obtained, and the overall improved interval is used as the interval used by the CLARANSI algorithm. Then, the CLARANSI algorithm is executed on the multi-information queue to cut the multi-information in the multi-information queue into more than one group.

7. The distributed photovoltaic group control method according to claim 6 is characterized in that: In Step 4, the attributes of each group are calculated, which include the mean of the output voltage value of each group and the number of multivariate information in the group, and the base index of each group attribute is calculated. The centralized outlier index of each group is calculated through the base index of each group attribute, and the multivariate information that generates outliers is selected. The distributed photovoltaic system corresponding to the multivariate information that generates outliers is a distributed photovoltaic system with hidden dangers of abnormal operation. In Step 4, the method for calculating the grouped concentrated outlier index includes: For a random group, the output voltage reference index of the group is obtained according to the output voltage value of the multivariate information in the group, the number reference index of the group is obtained according to the number of the multivariate information in the group, and the exponential subtraction between the output voltage reference index of the group and the number reference index of the group is obtained as the concentrated outlier index of the group; In Step 4, the calculation equation of the base index of the grouping attribute is: Here , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces; It is The average output voltage value of each group; and are the mean and standard deviation of the output voltage values ​​of all groups respectively; It is The number of multivariate information in each group; and are the mean and standard deviation of the number of multivariate information in all groups; In Step 4, after obtaining the benchmark index of the grouping attribute, the equation for calculating the concentrated outlier index of the grouping based on the benchmark index of the grouping attribute is: Here, It is The concentrated outlier index of the grouping; , They are The output voltage reference index and number reference index of each group; It is the sequence code of several groups cut into pieces; In Step 4, the concentrated outlier index of each group is obtained, and each group is arranged from high to low according to its concentrated outlier index. Among the arranged groups, the group with a pre-defined ratio is selected as the outlier group, and the multivariate information in the outlier group is regarded as the multivariate information generated by the outlier in the pre-defined time interval.

Images