CN111711417B

CN111711417B - Method for diagnosing abnormal power generation amount in photovoltaic system

Info

Publication number: CN111711417B
Application number: CN202010642831.4A
Authority: CN
Inventors: 张永
Original assignee: Fonrich Shanghai New Energy Technology Co ltd
Current assignee: Fonrich Shanghai New Energy Technology Co ltd
Priority date: 2020-07-06
Filing date: 2020-07-06
Publication date: 2021-07-13
Anticipated expiration: 2040-07-06
Also published as: CN111711417A

Abstract

The invention relates to a method for diagnosing abnormal power generation in a photovoltaic system. The string providing the direct current comprises a plurality of photovoltaic modules connected in series. A first voltage value and a first output power value of each photovoltaic module are collected at a first moment. And collecting a second voltage value and a second output power value of each photovoltaic module at a second moment. The power difference between the first and second output power values of each photovoltaic module is compared to the voltage difference between the first and second voltage values to obtain a data source. The data sources of all the photovoltaic modules are integrated into a data set, and abnormal data sources which are not in accordance with preset rules are selected from the data set, so that the photovoltaic modules with abnormal data sources are judged to be in a state of abnormal power generation amount.

Description

Method for diagnosing abnormal power generation amount in photovoltaic system

Technical Field

The invention mainly relates to the field of photovoltaic power generation, in particular to a method for diagnosing abnormal power generation in a photovoltaic system distributed with a large number of photovoltaic modules.

Background

The photovoltaic module, also called a solar panel, is a core part in a solar power generation system and also is the most important part in the solar power generation system, and has the function of converting solar energy into electric energy. The photovoltaic module needs to be detected, and the current detection means does not specify a standardized process and a detection method and often depends on experience accumulation. In addition, the photovoltaic module is used as the most core power generation component of the photovoltaic power station, and the quality directly influences the power generation efficiency of the power station. In the construction process of the photovoltaic power station, because the number of photovoltaic modules is large and the construction period is short, slight damage is easily caused in transportation, transportation and the like, the quality of the modules needs to be checked again in the process of entering, unloading and installing the modules, and the check is only surface detection.

The photovoltaic module works in outdoor severe environment for a long time, and the requirements on reliability and stability are particularly strict. More demanding requirements are placed on the production process of photovoltaic modules. In addition to visual inspection, new photovoltaic modules require extensive and more stringent inspection and certification before they are marketed. However, due to the reasons of materials, temperature and process, some problems still occur in some photovoltaic modules, such as uneven power generation of a cell, abnormal solder strip and the like, and once the problems occur in a certain cell, the normal power generation of the cell is seriously affected, and even the normal operation of the whole module is affected. Therefore, the abnormal position needs to be found in time when the photovoltaic module is inspected.

Photovoltaic power generation is a fluctuating and intermittent source of power that is uncontrollable relative to traditional sources of energy. The severe environment of the photovoltaic module causes the operation efficiency to be greatly influenced by the environment. Although the design life cycle of a photovoltaic module is about twenty-five years or so, the module efficiency and electrical component performance can decrease year by year. In fact, the photovoltaic module starts to decay after production. The attenuation is slow only when the package is not unpacked, the attenuation is accelerated after the package is unpacked and is irradiated by light, and besides the natural attenuation factors, the quality problems of components and inverters, various factors such as circuit layout, dust, series-parallel connection loss or cable loss and the like, such as the haze weather which is serious day by day, inevitably affect the efficiency and performance of photovoltaic power generation, and even cause abnormal attenuation. Abnormal attenuation can mean that the photovoltaic module can be out of order, and when the photovoltaic module is out of order, the problems that the secondary efficiency of the whole array is reduced due to the mismatching of a single module or a plurality of modules are easily caused. Therefore, if the power generation quality of the photovoltaic system can be evaluated and the attenuation and other problems of the photovoltaic module can be predicted, the fault prediction of the photovoltaic module can be realized, so that the photovoltaic module can be conveniently overhauled or replaced in advance to ensure the maximum benefit of the photovoltaic system, and the safety and the reliability of the photovoltaic system are ensured.

The existing detection method mainly adopts two technologies: one is electroluminescent detection technology. The electroluminescence detection mainly utilizes the electroluminescence principle to detect a darker area of the battery piece, namely, the current flowing in the area is small, but the whole detection process must be carried out in a full-black environment to avoid the influence of visible light on the detection, and the precision of the adopted electroluminescence detection camera has great influence on the detection result, so that the detection result is sometimes unsatisfactory. And secondly, the infrared thermal image detection technology is used for finding the position of the battery piece with the problem by detecting the heating condition of the battery piece. The temperature difference between slightly defective battery pieces is smaller, so that the infrared thermal image effect is not good, and the detection is easily influenced by heat transfer to generate misjudgment. It can be seen that, for detecting whether parts such as welding strips or battery pieces are abnormal or not, the existing detection methods have great defects, and whether the parts are abnormal or not can not be intuitively judged from data only through image judgment. Therefore, it is desirable to design a method for diagnosing abnormal power generation to detect a failure of a photovoltaic module to solve the above technical problems. With the continuous development of computer technology, especially the wide application of big data mining technology, the operation and maintenance level of the photovoltaic power station is greatly improved.

Disclosure of Invention

The present application relates to a method for diagnosing anomalies in the electrical energy production in a photovoltaic system, in which a string providing direct current comprises a plurality of photovoltaic modules connected in series, said method comprising:

collecting a first voltage value and a first output power value of each photovoltaic module at a first moment; and

collecting a second voltage value and a second output power value of each photovoltaic module at a second moment;

comparing the power difference value between the first output power value and the second output power value of each photovoltaic module with the voltage difference value between the first voltage value and the second voltage value to obtain a data source;

the data sources of all the photovoltaic modules are integrated into a data set, and abnormal data sources which do not accord with preset rules are selected from the data set, so that the photovoltaic modules with abnormal data sources are judged to be in a state of abnormal power generation.

The method described above, wherein:

synchronously acquiring respective first voltage values and first output power values of all the photovoltaic modules at a first moment;

synchronously acquiring respective second voltage values and second output power values of all the photovoltaic modules at a second moment;

and enabling the data sources of all the photovoltaic assemblies to be collected at the same time interval so as to form the data set in a data snapshot format.

The above method, wherein the state of abnormality in power generation amount includes:

the power generation capacity of the single photovoltaic assembly with the abnormal data source is lower than the power generation capacity of the other remaining single photovoltaic assemblies; or

The photovoltaic module with the data source with the abnormality does not work at the maximum power point.

In the method, an inverter is provided with direct current by a plurality of strings connected in parallel, the inverter performs inversion conversion from direct current to alternating current, and meanwhile, the inverter is also used for executing maximum power tracking so that the strings connected in parallel work at a maximum power point.

In the method, the mode of judging the fault rate between different groups of strings is as follows:

comparing the number of abnormal data sources of a plurality of parallel-connected string groups, the more the abnormal data sources are distributed, the higher the failure rate of the string groups exists.

In the method, the method for screening the group string with the fault comprises the following steps:

and adding the data sources of all the photovoltaic components of each group string to obtain a power point characteristic value object, wherein the power point characteristic value objects of all the group strings form an object set, and selecting an abnormal power point characteristic value object inconsistent with the rest other power point characteristic value objects from the object set, so that the group string with the abnormal power point characteristic value object is judged to be in a fault state.

In the above method, the method for determining whether the inverter operates at the maximum power point includes:

and adding the data sources of all the photovoltaic modules under the single inverter to obtain a sum value, comparing the sum value with a preset target value range, and judging that the inverter does not work at the maximum power point if the sum value is not in the preset target value range.

In the method, the normal distribution for all data sources is established by using the data set;

the preset rule is as follows: the distribution point of the data source in the normal distribution is within a preset confidence range; and

and defining the discrete data source which is not in the confidence range as the abnormal data source which does not conform to the preset rule.

periodically collecting the power variation and the voltage variation of each photovoltaic module, comparing the power variation and the voltage variation of each photovoltaic module, and taking the obtained ratio as a data source; or

Periodically collecting the power variation and the current variation of each photovoltaic module, comparing the power variation and the current variation of each photovoltaic module, and taking the obtained ratio as a data source;

and integrating the data sources of all the photovoltaic modules into a data set, selecting abnormal data sources inconsistent with the rest of other data sources from the data set, and judging that the photovoltaic modules with abnormal data sources are in the state of abnormal power generation.

In the above method, the method of selecting the abnormal data source is any one of an abnormal point detection method based on statistics, an abnormal point detection method based on distance, an abnormal point detection method based on deviation, and an abnormal point detection method based on density.

In the method, a plurality of strings connected in parallel provide direct current for an inverter, the inverter performs inversion conversion from direct current to alternating current, and the inverter further performs maximum power tracking to enable the strings to work at a maximum power point.

The above method, wherein the state of abnormality in power generation amount includes: the photovoltaic module with the data source with the abnormality does not work at the maximum power point.

The present application relates to a method for diagnosing an abnormal amount of power generation in a photovoltaic system, wherein a string for supplying direct current comprises a plurality of photovoltaic modules connected in series, a plurality of strings connected in parallel supply direct current to an inverter, the inverter performs an inversion conversion of direct current to alternating current, and the inverter further performs maximum power tracking to operate the string at a maximum power point, the method comprising:

calculating the power variation and the voltage variation of each photovoltaic module once every a set time period, comparing the power variation and the voltage variation of each photovoltaic module, and taking the obtained ratio as a data source; or

Calculating the power variation and the current variation of each photovoltaic module once every a set time interval, comparing the power variation and the current variation of each photovoltaic module, and taking the obtained ratio as a data source;

the data sources of all the photovoltaic modules are integrated into one data set, and a normal distribution probability model for all the data sources is established by utilizing the data set;

defining the data source with the distribution point within a preset confidence range in normal distribution as a normal data source;

defining a data source with a distribution point not within a preset confidence range in normal distribution as an abnormal data source;

therefore, the photovoltaic module with the abnormal data source is judged to be in the state of abnormal power generation.

The present application relates to a method for diagnosing anomalies in the electric energy production in a photovoltaic system, characterized in that a string providing direct current comprises a plurality of photovoltaic modules connected in series, said method comprising:

collecting a first current value and a first output power value of each photovoltaic module at a first moment; and

collecting a second current value and a second output power value of each photovoltaic module at a second moment;

comparing the power difference value between the first output power value and the second output power value of each photovoltaic module with the current difference value between the first current value and the second current value to obtain a data source;

According to the scheme disclosed in the context of the present application, the main idea of the present application is to replace the traditional appearance detection or the traditional physical detection with a completely new detection means of big data mining. Recessive wound flaws that are difficult to find by naked eyes in the traditional appearance detection process can be reflected in the aspect of data mining of photovoltaic modules. The traditional physical detection such as electroluminescence detection or infrared thermography detection and the like can also reflect the data mining aspect of the photovoltaic module. The novel detection means of this application belongs to on-line measuring completely and also need not to dismantle any photovoltaic module individuality from photovoltaic system, and this intelligence fortune dimension is photovoltaic electric power fortune dimension and combines the novel power management scheme that data mining and produced, has constructed new fortune dimension service support system and has effectually promoted photovoltaic power plant operation's security and high efficiency and economic nature, provides intelligent analysis and accurate guidance for photovoltaic power plant's safe operation, failure diagnosis, state early warning etc. lets photovoltaic power plant's operation safe and reliable more.

Drawings

To make the above objects, features and advantages more comprehensible, embodiments accompanied with figures are described in detail below, and features and advantages of the present invention will become apparent upon reading the following detailed description and upon reference to the following figures.

Fig. 1 shows photovoltaic modules connected in series to form strings and groups connected in parallel to provide dc power.

Fig. 2 is a photovoltaic inverter with maximum power tracking to set the string at the maximum power point.

Fig. 3 shows that the pv inverter dynamically performs maximum power point tracking by actively adjusting the bus voltage.

Fig. 4 is a diagram of both components operating at the maximum power point and components offset from the maximum power point.

Fig. 5 is a graph in which the power variation amount and the voltage variation amount of the photovoltaic module are calculated once every one period of time.

FIG. 6 is a probability distribution curve of a data source of a photovoltaic module under a normal distribution or Gaussian distribution model.

Fig. 7 is a graph in which the power variation amount and the current variation amount of the photovoltaic module are calculated once every one period of time.

Detailed Description

The invention will be more fully described with reference to the following examples. The solutions obtained by a person skilled in the art without making any inventive step are within the scope of protection of the present invention.

In the field of photovoltaic power generation, a photovoltaic module or a photovoltaic cell is a core component of power generation. The solar cell panel is divided into a monocrystalline silicon solar cell, a polycrystalline silicon solar cell, an amorphous silicon solar cell and the like in the direction of mainstream technology, the required service life of the silicon cell is as long as more than twenty years, and the output characteristic of the cell is essential to be monitored for durability. Many internal and external factors contribute to inefficient power generation of photovoltaic modules: conversion efficiency is reduced due to manufacturing differences or installation differences between the photovoltaic modules themselves or shading or maximum power tracking adaptation. Taking the shielding as an example, if a part of the photovoltaic modules are shielded by clouds, buildings, tree shadows, dirt and the like, the part of the photovoltaic modules can be changed into a load by a power supply and does not generate electric energy any more, the temperature of the photovoltaic modules at a local position with a serious hot spot effect is higher, and some photovoltaic modules can exceed hundreds of degrees centigrade to cause permanent damage such as burning or dark spots, welding spot melting, packaging material aging, glass explosion, corrosion and the like, and great potential hazards are caused to the long-term safety and reliability of the photovoltaic modules. The problems to be solved by the photovoltaic power generation system are as follows: the working state of each installed photovoltaic cell panel can be observed in real time or intermittently, early warning can be carried out on various abnormal conditions such as potential power generation faults of the photovoltaic cells, and the early warning is particularly important for taking local maintenance or active replacement or other emergency measures for the abnormal photovoltaic cells.

Referring to fig. 1, a photovoltaic module array is the basis for the conversion of light energy to electrical energy in a photovoltaic power generation system. A string of battery cells mounted in a photovoltaic module array is shown. Regarding the battery string ST: each battery string is formed by connecting a plurality of photovoltaic modules which are connected in series. The photovoltaic module can also be replaced by a direct current power supply such as a fuel cell or a chemical battery. A plurality of different battery strings are connected in parallel: although each battery string is composed of a plurality of photovoltaic modules and the plurality of photovoltaic modules inside are connected in series, the different battery strings ST are connected in parallel with each other and collectively supply electric power to an energy collecting device such as the photovoltaic inverter 200.

Referring to fig. 1, in a single battery string ST: voltage V output by first photovoltaic module PV1₁Plus the voltage V output by the second photovoltaic module PV2₂Plus a third photovoltaic groupVoltage V output by member PV2₃… …, and so on until the voltage V output by the nth pv module PVN is added up_NSo that the operation result of the string-level voltage of the single battery string ST is equal to V₁+V₂+……V_N. The positive integer N is greater than 1. In the tandem type photovoltaic array, the application takes the tandem type multi-stage photovoltaic modules PV1-PVN as an example to take the respective output voltages V thereof₁To V_NAfter addition, the resulting voltage cascade with the higher potential, i.e. the bus voltage V_BUSAnd the output power is provided for the inverter 200, and the inverter 200 performs inversion from direct current to alternating current after converging the respective output power of the serially connected multistage photovoltaic modules. Capacitors with large capacity are often connected between the dc buses providing the dc power supply for the inverter 200, and the large capacitors between the dc buses must also perform decoupling between the relatively constant input power and the fluctuating output power of the inverter in the photovoltaic power generation system.

Referring to fig. 1, photovoltaic inverter 200 has maximum power point tracking MPPT functionality. Photovoltaic power generation is greatly influenced by temperature and irradiance, and in order to obtain more electric energy under the same condition and improve the operation efficiency of a system, the tracking of the maximum power point of a photovoltaic cell becomes a long-standing problem in the development of the photovoltaic industry. Early researches on the maximum power point tracking technology of a photovoltaic array mainly comprise a constant voltage tracking method, a photovoltaic array combination method and an actual measurement method. The constant voltage tracking method is actually equivalent to voltage stabilization control, and does not achieve the purpose of maximum power point tracking. The photovoltaic array combination method is used for adjusting the number of series-parallel connection of photovoltaic arrays according to different loads, and has no real-time property. The actual measurement method is to use an additional photovoltaic array module to establish a reference model of the photovoltaic array at a certain sunshine amount and temperature, and the method does not consider the real-time shading condition and the difference of each solar panel. At present, the maximum power tracking method of the photovoltaic array is mainly divided into a method based on a mathematical model, a self-optimizing method based on disturbance and a method based on an intelligent technology. The method based on the mathematical model is to construct a solving method and a photovoltaic array characteristic curve by taking the optimized mathematical model as a starting point, so that the maximum power output of the photovoltaic array is obtained, and therefore, the equivalent circuit model of the photovoltaic cell and the correctness of various parameters need to be considered emphatically.

Referring to fig. 1, the principle and features of the MPPT method commonly used in the power optimizer, such as the early output power control for the photovoltaic module, mainly use the Constant Voltage Tracking method, which ignores the influence of temperature on the open-circuit Voltage of the solar cell, so the open-circuit Voltage method and the short-circuit current method are proposed, and their common property is basically very similar to the maximum power point. In order to more accurately capture the maximum power point, a disturbance observation method, a duty ratio disturbance method, a conductance increment method and the like are proposed. The disturbance observation method adopts the working principle that the current array output power is measured, then a small voltage component disturbance is added to the original output voltage, the output power can be changed, the changed power is measured, the power before and after the change is compared, the power change direction can be known, if the power is increased, the original disturbance is continuously used, and if the power is reduced, the original disturbance direction is changed. The working principle of duty cycle disturbance is as follows: the interface between the photovoltaic array and the load usually adopts a voltage converter controlled by a pulse width modulation signal, and the input and output relation of the converter is adjusted by adjusting the duty ratio of the pulse width modulation signal, so that the function of impedance matching is realized, and therefore, the magnitude of the duty ratio substantially determines the magnitude of the output power of the photovoltaic cell. The incremental conductance method is a special way with the disturbance observation method, the biggest difference is only in the logic judgment formula and the measurement parameters, although the incremental conductance method still changes the output voltage of the photovoltaic cell to reach the maximum power point, the oscillation phenomenon near the maximum power point is reduced by modifying the logic judgment formula, so that the incremental conductance method is more suitable for the climate conditions with the instantaneous change of the sunlight intensity and the temperature. The actual measurement method, the fuzzy logic method, the power mathematical model method, the intermittent scanning tracking method, the optimal gradient method, the three-point gravity center comparison method and the like belong to the most power point tracking schemes which are not commonly used, so that the fact that the MPPT algorithm is diversified can be known, and the method is not repeated in the application.

Referring to fig. 2, the power voltage P of a string ST with photovoltaic modules_ST-V_STIn the graph, any cluster has a unique maximum output power point P under the same environmental conditions_MX. At the maximum power point P_MXOn the left, the output power of the string, shown as a curve, shows a substantially linear increasing trend as the output voltage of the string increases. The output power of the string after the maximum power point is reached, which is represented by the curve, will drop rapidly, with the rate of the drop being much greater than the rate of the rise. Maximum power point tracking requires stabilizing the output voltage of the string at the maximum power point P_MXCorresponding output voltage V_MPNearby. It must be appreciated that maximum power point tracking for a string does not mean that individual photovoltaic modules within the string also operate at the maximum power point.

Referring to fig. 3, in the single battery string ST: the first photovoltaic module PV1 has an output voltage V₁And the second photovoltaic module PV2 outputs a voltage V₂And the third photovoltaic module PV2 outputs a voltage V₃… …, and so on until the Nth photovoltaic module PVN output voltage at the end is V_N. The external real-time temperature, the current light irradiation intensity, whether the light irradiation intensity is shielded or not and other environmental factors directly influence the output voltage of the photovoltaic modules, and even if the output voltages of different photovoltaic modules in the completely same string are different. To output a voltage V₁And output power V_NThe curves in the figures are illustrated as representative examples: voltage V caused by external factors₁To V_NEtc., it is necessary to introduce the maximum power point tracking function of the inverter 200. Inverter 200 may be implemented by varying voltage V during the step of maximum power point tracking_BUSTo detect the maximum power point. The inverter 200 actively adjusts the bus voltage low to perform power tracking, for example, during the first period from time T1 to T2, and the inverter 200 actively adjusts the bus voltage high to perform power tracking, for example, during the second period from time T2 to T3. In an alternative example, the inverter periodically adjusts the voltage value of the bus voltage, i.e., the voltage in the cascade when performing the maximum power point tracking. Each time the inverter regulates the voltage value of the bus voltage, the output voltage of each photovoltaic module in the string ST is more or lessThe voltage share of the output voltage of the respective photovoltaic module, which is distributed from the busbar voltage, is changed in dependence on the output power of the respective photovoltaic module. Although the regulation action of the inverter for performing voltage regulation on the cascade voltage guides the string to find the maximum power point, the difficulty of finding the maximum power point by each photovoltaic module is increased, and the difficulty of analyzing whether each individual photovoltaic module works at the maximum power point is increased because the output voltage of each individual photovoltaic module oscillates along with the regulation of the bus voltage.

Referring to fig. 4, maximum power point tracking for the string ST does not mean that the individual photovoltaic modules inside the string ST also operate at the maximum power point. The figure shows a characteristic graph of current versus voltage for each of the photovoltaic modules PV1-PVN connected in series within the string ST. The characteristic diagram of the first photovoltaic module PV1 shows the current I₁To voltage V₁And a second photovoltaic module PV2 is characterized by a current I₂To voltage V₂And the third photovoltaic module PV3 is characterized by a current I₃To voltage V₃… …, and so on until the Nth characteristic diagram of the last PV module PVN also shows the current I_NTo voltage V_N. The series connection relation determines the current I₁To current I_NAre equal. The signature graph for any one photovoltaic module shows that the current decreases as the output voltage increases. The output power of each photovoltaic module is equal to the product of the current and the voltage and the output power varies with the variation of the output voltage. The purpose of so-called MPPT, maximum power point tracking, is to find the maximum power point at which the system can operate to maximize the power drawn from the components.

Referring to fig. 4, the maximum power point tracking technique is discussed in the article entitled "comparison of several photovoltaic array maximum power point tracking techniques" published in the IEEE energy conversion journal by t.espam and p.l.chapman, the entire disclosure and discussion of which is considered to be prior art and reference of the present application. The PV modules PV1-PVN are connected in series to the inverter that performs maximum power point tracking, and the MPPT function must select a single power point, which is approximately the average of the maximum power points of the individual PV modules PV1-PVN for maximum power point tracking of the string ST. Referring to the characteristic graph of current versus voltage of each photovoltaic module: it is possible that the power point at which MPPT is implemented for the string ST is such that the maximum power point of itself is only approached at a few photovoltaic modules. For example, PV3 operates at its maximum power point, PV PVN is relatively close to its maximum power point, and the remaining other PV modules, such as PV1 and PV2, do not operate at their maximum power points. The power point achieved for a string can only bring a few photovoltaic modules close to their own maximum power point, even worse, none of them works at their own maximum power point. Needless to say, the maximum power point tracked by the inverter can only be considered as the average value of the MPP of the series photovoltaic modules, and the thick black point in the characteristic graph of current versus voltage of each photovoltaic module represents the actual maximum power point of the photovoltaic module, under which condition the respective photovoltaic module cannot operate at maximum efficiency.

Referring to fig. 4, even though the maximum power point tracking for the string ST can only make a few photovoltaic modules close to their maximum power points, the power difference between the modules is not known from the appearance of the photovoltaic modules, and it is not practical to detect the power of each photovoltaic module in a power generation system with a large number of photovoltaic modules. Therefore, it is necessary to detect the abnormal power generation condition of the photovoltaic module by means of data mining to determine whether the photovoltaic module has a fault. If the power generation quality of the photovoltaic system can be evaluated and predicted, the fault prediction of the photovoltaic module can be realized, so that the maintenance or replacement preparation of the photovoltaic module is implemented in advance, and the operation safety of the photovoltaic system is ensured.

Referring to fig. 5, a first voltage value and a first output power value of each photovoltaic module are collected at a first TIME1 and a second voltage value and a second output power value of each photovoltaic module are collected at a second TIME 1. The description is temporarily illustrated by taking as an example the totaling N photovoltaic modules PV1-PVN among a single string: collecting a first electricity of a first photovoltaic module PV1 at a first TIME instant TIME1Pressure value V₁-1 and a first output power value P₁-1 and a second TIME instant TIME2 for acquiring a second voltage value V of the first photovoltaic module PV1₁-2 and a second output power value P₁-2. The first TIME point 1 detects a first voltage value V of a second photovoltaic module PV2₂-1 and a first output power value P₂-1 and a second TIME instant TIME2 for collecting a second voltage value V of a second photovoltaic module PV2₂-2 and a second output power value P₂-2. The first TIME TIME1 collects a first voltage value V of a third photovoltaic module PV3₃-1 and a first output power value P₃-1 and a second TIME instant TIME2 of collecting a second voltage value V of a third photovoltaic module PV3₃-2 and a second output power value P₃-2. According to the same principle, the first voltage value V of the nth photovoltaic module PVN can be collected at the first TIME1 in the string by analogy_N-1 and a first output power value P_N-1 and collection of a second voltage value V of the nth photovoltaic module PVN at TIME2_N-2 and a second output power value P_N-2. It has been clarified that the output power of each photovoltaic module is equal to the product of the current and the voltage, and the current of the photovoltaic module can be easily collected by the current detection module of the prior art, the voltage of the photovoltaic module can be easily collected by the voltage detection module of the prior art, and the voltage value and the output power value of the photovoltaic module can be collected by the prior art.

Referring to fig. 5, the power difference between the first and second output power values of the first photovoltaic module is compared to the voltage difference between the first and second voltage values to obtain a first data source. For example, the power difference (i.e., P power difference) between the first and second output power values of the first photovoltaic module PV1₁-2 minus P₁-1) and the voltage difference between the first and second voltage values (voltage difference, i.e. V)₁-2 minus V₁-1) comparing, resulting in data source Δ P1/Δ V1. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P1 ═ P₁-2 minus P₁-1。ΔV1＝V₁-2 minus V₁-1。

Referring to FIG. 5, the power between the first and second output power values of the second photovoltaic module is measuredThe difference is compared to a voltage difference between the first and second voltage values to obtain a first data source. For example, the power difference (i.e., P power difference) between the first and second output power values of the second photovoltaic module PV2₂-2 minus P₂-1) and the voltage difference between the first and second voltage values (voltage difference, i.e. V)₂-2 minus V₂-1) comparing, resulting in data source Δ P2/Δ V2. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P2 ═ P₂-2 minus P₂-1。ΔV2＝V₂-2 minus V₂-1。

Referring to fig. 5, the power difference between the first and second output power values of the third photovoltaic module is compared to the voltage difference between the first and second voltage values to obtain a first data source. For example, the power difference (i.e., P power difference) between the first and second output power values of the third photovoltaic module PV3₃-2 minus P₃-1) and the voltage difference between the first and second voltage values (voltage difference, i.e. V)₃-2 minus V₃-1) comparing, resulting in data source Δ P3/Δ V3. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P3 ═ P₃-2 minus P₃-1。ΔV3＝V₃-2 minus V₃-1。

Referring to fig. 5, the power difference between the first and second output power values of the nth photovoltaic module is compared to the voltage difference between the first and second voltage values to obtain a first data source. For example, the power difference (i.e., P) between the first and second output power values of the Nth PV module PVN_N-2 minus P_N-1) and the voltage difference between the first and second voltage values (voltage difference, i.e. V)_N-2 minus V_N-1) comparing, resulting in a data source Δ PN/Δ VN. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ PN ═ P_N-2 minus P_N-1。ΔVN＝V_N-2 minus V_N-1。

Referring to fig. 5, taking N photovoltaic modules PV1-PVN of a single string as an example, wherein data sources of all the photovoltaic modules are integrated into one data set { Δ P1/Δ V1, Δ P2/Δ V2, Δ P3/Δ V3 … … Δ PN/Δ VN }, it is required to select an abnormal data source which does not meet a preset rule from the data sets { Δ P1/Δ V1, Δ P2/Δ V2, Δ P3/Δ V3 … … Δ PN/Δ VN }, and determine that the photovoltaic module whose data source has an abnormal power generation amount is in an abnormal state.

Referring to fig. 5, the data sources of the N PV modules PV1-PVN may be grouped together in a single string as a data source for big data analysis, or the data sources of each of the different strings may be grouped together as a larger data set for big data analysis. The data sets Δ P1/Δ V1- Δ PN/Δ VN of the first group of strings may be considered as the first data source set and the data sets Δ P1/Δ V1- Δ PN/Δ VN of the second group of strings may be considered as the second data source set, further even the data sets Δ P1/Δ V1- Δ PN/Δ VN of the third group of strings may be considered as the third data source set and the data sets Δ P1/Δ V1- Δ PN/Δ VN of the fourth group of strings may be considered as the fourth data source set, if there are more group strings that can also provide data source sets of other group strings, the data sets Δ P1/Δ V1- Δ PN/Δ VN of the kth group of strings may be considered as the kth data source set and the positive integer K is greater than 1. According to the analysis principle: and a first data source set of the first group of strings, a second data source set of the second group of strings, a third data source set of the third group of strings, a fourth data source set of the fourth group of strings, … … and so on until a Kth data source set of the Kth group of strings and the like are collected to be used as a larger data set, and abnormal data sources which do not accord with preset rules need to be selected from the larger data set, so that the photovoltaic modules with abnormal data sources are judged to be in the state of abnormal power generation amount. The larger data set in this example is based on K x N data, while the single string is based on N data, in fact the more samples the more accurate results. Note that the sets of strings from which the dataset samples are taken may or may not have a parallel connection relationship between them.

Referring to fig. 5, on the basis of big data analysis, it can be generally determined that the photovoltaic module with the abnormal data source is in the state of abnormal power generation. For an abnormal photovoltaic module, the state of abnormal power generation amount includes: the power generation amount of the single photovoltaic module with the abnormal data source is lower than that of the other remaining single photovoltaic modules, namely the power generation amount of the single photovoltaic module with the abnormal data source is lower than that of the single photovoltaic module without the abnormal data source. If the maximum power point tracking technology is implemented, the photovoltaic module is required to work at the maximum power point of the photovoltaic module, so the abnormal state of the power generation amount also comprises the condition that the photovoltaic module with the abnormal data source does not work at the maximum power point.

Referring to FIG. 5, data mining is the process of extracting potential, implicit, valuable knowledge or patterns or rules from large-scale data sets. The patterns mined from large-scale datasets can be generally classified into five categories: association rules, classification and prediction, clustering, evolution analysis, anomaly detection, and the like. The mining of the abnormal point data comprises two parts of abnormal point data detection and abnormal point data analysis. The analysis of the abnormal point data, namely the abnormal data source which does not conform to the preset rule needs to be researched by combining with relevant knowledge such as industry background knowledge, field knowledge and the like. The application focuses on the detection problem of abnormal point data, namely an abnormal data source, which is the most critical part in abnormal point data mining. Outlier data is data that is inconsistent with the general behavior or model of the data, which is distinctive data in the data set that is not a random bias but results from a completely different mechanism. The abnormal point data mining has wide application, such as battery detection, and abnormal point detection is used for detecting unusual photovoltaic modules. Through research on the data, abnormal power generation behaviors and actual operation modes of the photovoltaic module are found, and therefore the abnormal data mining function is achieved.

Referring to fig. 5, an article entitled "analysis of anomaly detection algorithms" was published by authors, listeria and listeria et al, in computer engineering 2002.028(006), and Hawkins gives an essential definition of an anomaly: anomalies are data that are distinctive in a data set, leaving one to suspect that the data is not a random deviation, but arises from a completely different mechanism. And the definition of the clustering algorithm to the anomaly: the anomaly is background noise embedded in the cluster. Definition of anomalies by anomaly detection algorithms: an anomaly is a point that does not belong to either a cluster or background noise, and the behavior of the anomaly point is completely different from the behavior of normal data. It is generally believed that the following data objects may exist: such exception objects (outliers) are referred to as exception points and they are different or inconsistent with other parts of the data, not conforming to the general model of the data. The source of the exception data belongs to an exception point. The mining of outlier data by the present application can be described as follows, given a collection containing data sources or data objects, the goal is: finding data objects that are significantly different or abnormal or inconsistent from the remaining data results in a photovoltaic module that behaves abnormally. The prior art anomaly point data detection, namely inconsistency detection, comprises the following steps: (1) statistical-based methods use inconsistency tests to determine outlier data according to known data distributions or probabilistic models. (2) The distance-based method calculates the distance between data to obtain the abnormal point data. (3) The bias-based approach determines the existence of outlier data through the impact and variation of the entire data set from the whereabouts of a record in the data. (4) The density-based approach detects locally anomalous data points by calculating local anomaly factors for the data points in the data set. (5) And (3) anomaly detection of the high-dimensional data, namely mapping the high-dimensional data to a low-dimensional subspace, and determining anomalous data according to the sparsity degree of the subspace mapping data. In other words, the abnormal data source inconsistent with the rest of the data sources needs to be selected from the data set, and the selection method can be selected from one of an abnormal point detection method based on statistics, an abnormal point detection method based on distance, an abnormal point detection method based on deviation and an abnormal point detection method based on density.

Referring to fig. 5, taking N photovoltaic modules PV1-PVN of a single string as an example, wherein data sources of all the photovoltaic modules are integrated into one data set { Δ P1/Δ V1, Δ P2/Δ V2, Δ P3/Δ V3 … … Δ PN/Δ VN }, it is necessary to select an abnormal data source inconsistent with the rest of the data sources from the data sets { Δ P1/Δ V1, Δ P2/Δ V2, Δ P3/Δ V3 … … Δ PN/Δ VN }. The detection method may be any one of a statistical-based abnormal point detection method, a distance-based abnormal point detection method, a deviation-based abnormal point detection method, and a density-based abnormal point detection method.

Referring to fig. 5, the data sets Δ P1/Δ V1- Δ PN/Δ VN of the first group of strings may also be considered as the first data source set and the data sets Δ P1/Δ V1- Δ PN/Δ VN of the second group of strings may be considered as the second data source set, further, even the data sets Δ P1/Δ V1- Δ PN/Δ VN of the third group of strings may be considered as the third data source set and the data sets Δ P1/Δ V1- Δ VN of the fourth group of strings may be considered as the fourth data source set, if there are more group strings that can also provide data source sets of other group strings, the data sets Δ P1/Δ V1- Δ PN/Δ VN of the kth group of strings may be considered as the kth data source set and the positive integer K is greater than 1. According to the analysis principle: the first data source set of the first group of strings, the second data source set of the second group of strings, the third data source set of the third group of strings, the fourth data source set of the fourth group of strings, … … and so on until the Kth data source set of the Kth group of strings and the like are collected to be a larger data set, and abnormal data sources inconsistent with the rest of other data sources still need to be selected from the larger data set. Similarly, the abnormal point detection method may be any one of a statistical-based abnormal point detection method, a distance-based abnormal point detection method, a deviation-based abnormal point detection method, and a density-based abnormal point detection method.

Referring to fig. 6, normal or gaussian distributions are models of quantitative phenomena in natural science and behavioral science. Normal distributions occur in many statistical domains: for example, the mean of the sampling distribution is approximately normal, even though the original population distribution of the sampled samples does not follow a normal distribution. Normal distributions are the most widely used type of distribution in statistics and statistical testing. The normal distribution in probability theory is the limit distribution of several continuous and discrete distributions, and the function of the normal distribution is as follows:

the random variable x obeys a probability distribution with a position parameter of mu and a scale parameter of sigma, the random variable x is generally called a normal random variable, and the distribution obeyed by the normal random variable is normal distribution or Gaussian distribution. The location parameter is essentially a mean or average and the scale parameter is essentially a standard deviation. Characteristic of normal distribution: (1) concentration, the highest peak of the curve is located at the midpoint, and the position is the position of the mean. (2) And the symmetry is that the normal distribution curve is symmetrical left and right by taking the position of the mean number as the center, and two sections of the curve are close to the horizontal axis in a wireless manner. (3) The normal distribution curve uniformly descends towards the left side and the right side by taking the position of the mean number as the center. (4) The area is constant, and the area between the curve and the transverse axis is equal to 1.

Referring to fig. 6, the position parameter μ is a mean value and the scale parameter σ is a standard deviation. The position parameter μ determines the position of the normal distribution and the closer to μ the higher the probability of being taken, and vice versa. The scale parameter σ describes the degree of dispersion of the normal distribution and the larger σ the flatter the data distribution, and the smaller σ the steeper the data distribution. In the application, a preset mean value mu and a preset standard deviation sigma under normal distribution can be designed in advance, data sources of all photovoltaic modules are integrated into one data set, and the data source serving as a normal random variable obeys a normal distribution probability model under the conditions of the preset mean value and the preset standard deviation. And processing the detection of the abnormal point data source by using a statistical method, wherein the statistical method assumes a normal distribution probability model for a given data source set, and then determines the abnormal point data source by adopting an inconsistency test according to the normal distribution probability model. Regarding the inconsistency test, since the data set model parameters such as the preset mean and the preset standard deviation have been designed in advance, the abnormal data source can be easily analyzed and tested.

Referring to fig. 6, in a normal distribution graph, the scale on the horizontal axis is normalized to the midpoint, i.e., μ, meaning that the mean is the center of the curve and that this is the highest point of the curve, since most points or data sources are located near the mean. The scale on the horizontal axis to the left of the midpoint is negative-3 σ to 0 and the scale on the horizontal axis to the right of the midpoint is positive 3 σ to 0, meaning that if the midpoint is taken as the standard, the data source that is off the midpoint is different from the midpoint standard. The left side of the horizontal axis-1 σ is the data source range one standard deviation less than the midpoint criterion, and the right side of the horizontal axis-1 σ is the data source range one standard deviation more than the midpoint criterion. The left-2 σ of the horizontal axis is the data source range two standard deviations less than the midpoint criterion, and the right-2 σ is the data source range two standard deviations more than the midpoint criterion. The-3 σ on the left side of the horizontal axis is the data source range three standard deviations less than the midpoint standard and the 3 σ on the right side of the horizontal axis is the data source range three standard deviations more than the midpoint standard.

Referring to FIG. 6, in an alternative but not required example of a normal distribution plot, the percentage of data sources within a numerical range (+ -1 σ) from the mean or standard deviation is 68.2% of all data sources. The ratio of data sources within a numerical range (+ -2 sigma) from the mean or two standard deviations is 95.5% of all data sources. The percentage of data sources within a numerical range (+ -3 sigma) from the mean or standard three standard deviations is 99.7% of all data sources. This is merely a rule of thumb so that a normal distribution can be established for all data sources using a data set, and then anomalous data sources that do not meet a preset rule are selected from the data set. The so-called preset rules may be set as: the distribution point of the data source in a normal distribution must be within a predetermined confidence range. The confidence range is also called confidence range, and normal random variables such as data source Δ PN/Δ VN are generally unlikely to reach 100% confidence or confidence. The predetermined confidence range is, for example, a numerical range (± 1 σ) within one standard deviation from the mean or standard on the horizontal axis, or the predetermined confidence range is, for example, a numerical range (± 2 σ) within two standard deviations from the mean or standard on the horizontal axis, or the predetermined confidence range is, for example, a numerical range (± 3 σ) within three standard deviations from the mean or standard on the horizontal axis, or the predetermined confidence range is another range. This allows an easy estimation of the variability of the variable data source Δ PN/Δ VN and gives a corresponding confidence level, how many of its possible values are, for example, in a normal distribution curve, a probability of 68.2% for a data source Δ PN/Δ VN of ± 2 σ, and for example a probability of 99.7% for the same variable data source Δ PN/Δ VN of ± 3 σ. And the data source with the distribution point in the normal distribution probability model within a preset confidence range is defined as a normal data source, and the discrete data source which is not within the confidence range is defined as an abnormal data source which does not accord with the preset rule. For example, if data sources within a predetermined confidence range (± 3 σ) are defined as normal data sources and discrete data not within the confidence range (± 3 σ) are defined as abnormal data sources, then data sources more than three standard deviations from the mean value belong to the abnormal data sources, and a range distributed further from the-3 σ point, i.e., a range on the left side of-3 σ, and a range distributed further from the 3 σ point, i.e., a range on the right side of 3 σ, belong to the abnormal data sources.

Referring to FIG. 6, data sources that are within a predetermined confidence range in a normal distribution are defined as normal data sources, and data sources that are not within a predetermined confidence range in a normal distribution are defined as abnormal data sources. The Confidence range or Confidence interval refers to an estimation interval of the overall parameter constructed by the sample statistic, and the Confidence interval (Confidence interval) of a probability sample in statistics is an interval estimation of a certain overall parameter of the sample. The confidence interval represents the degree to which the true value of a parameter, such as a data source, falls around the measurement with a certain probability, which gives the confidence level, i.e. the required confidence probability, of the measured value of the measured parameter, such as a data source.

Referring to fig. 5, a first TIME1 collects a first voltage value V of the nth photovoltaic module PVN_N-1 and a first output power value P_N-1 and collection of a second voltage value V of the nth photovoltaic module PVN at TIME2_N-2 and a second output power value P_N-2. The TIME width between the first TIME instant TIME1 and the second TIME instant TIME2 in many scenarios is allowed to be somewhat shorter than the period during which the inverter 200 regulates the bus voltage, or the TIME width is allowed to be somewhat shorter than the regulation period, such as T1-T2 or T2-T3, during which the voltage regulation is performed on the string-level voltage in order to find the maximum power point of the string. In more precise data mining, the time width, which is actually the time duration between the first time and the second time, can be set to be slightly longer than the adjustment period for adjusting the voltage value of the string voltage. The output voltage of each photovoltaic module follows the cascade voltageThe regulation is changed along with the regulation, the regulation of the cascade voltage is from the power optimization purpose, the change of the output voltage of the photovoltaic module caused by the voltage regulation action of the cascade voltage is different from the change of the output voltage of the photovoltaic module caused by environmental factors such as temperature or illumination radiance, the former is caused by an inverter, and the latter is caused by irresistible natural factors. Under the influence of different interference sources such as natural environment factors, active voltage regulation behaviors and the like, the data source obtained by calculation and the mining result of big data of the photovoltaic module are closer to the real running condition of the photovoltaic module, so if the set time width is longer than the regulation period of the voltage value of the voltage of the inverter regulation cascade, the potential power generation fault of the photovoltaic cell can be embodied. The reason for this is that the voltage regulation action performed by the inverter on the cascade voltage takes place between the two data acquisition instants.

Referring to fig. 1, the data sources of N PV1-PVN may be grouped together in a single string as a data source for big data analysis, or the data sources of different strings may be grouped together as a larger data set for big data analysis. For example, the first data source set of the first group string, the second data source set of the second group string, and the third data source set of the third group string, the fourth data source set of the fourth group string are described above, … … and so on until the so-called Kth data source set of the Kth group string, etc., are collected as a larger data set. The manner of determining the failure rate between different strings may be: comparing the number of the abnormal data sources of each group string, the failure rate of the group string with more distribution of the abnormal data sources is higher, and at the moment, the group strings are allowed to have a parallel connection relationship or do not have the parallel connection relationship. For example, the number of abnormal data sources in the first data source set of the first group of strings is greater than the number of abnormal data sources in the second data source set of the second group of strings, and the first group of strings with more distributed abnormal data sources has a higher failure rate than the second group of strings without any doubt. For example, the number of abnormal data sources in the fourth data source set of the fourth group string is less than that in the kth data source set of the kth group string, and the failure rate of the fourth group string with less abnormal data source distribution is lower than that of the kth group string without any doubt.

Referring to fig. 5, in an alternative example, the power variation and the voltage variation of each photovoltaic module may be periodically collected and compared with the voltage variation to obtain a data source. For example, the power variation (such as P) of the PVN of the Nth photovoltaic module is periodically collected_N-2 minus P_N-1, and the variation of the acquisition voltage (e.g. V)_N-2 minus V_N-1 this amount of change). And comparing the power variation delta PN of the photovoltaic module PVN with the voltage variation delta VN to obtain a data source delta PN/delta VN. The TIME width between the first TIME1 and the second TIME2 corresponds to a TIME period for collecting the power variation and the voltage variation of each photovoltaic module. The power variation and the voltage variation of the photovoltaic module are periodically collected, which means that the power variation and the voltage variation of the photovoltaic module are continuously and cyclically captured by taking the time width as a period. For example, in a first period, acquiring the power variation and the voltage variation of each photovoltaic module to obtain the data source of each photovoltaic module and further obtain the data set of the data source, then in a second period, acquiring the power variation and the voltage variation of each photovoltaic module to obtain the data source of each photovoltaic module and further obtain the data set of the data source, and then in a third period, acquiring the power variation and the voltage variation of each photovoltaic module to obtain the data source of each photovoltaic module and further obtain the data set of the data source, and repeating the steps to continuously and continuously detect the photovoltaic modules. The length of TIME of each cycle period is equal to the width of TIME between TIME1 and TIME 2.

Referring to fig. 5, in an alternative example, the power variation and the voltage variation of each photovoltaic module may be collected non-periodically and compared with the voltage variation to obtain a data source. For example, randomly collecting the power variation (such as P) of the PVN of the Nth photovoltaic module_N-2 minus P_N-1, and the variation of the acquisition voltage (e.g. V)_N-2 minusGo V_N-1 this amount of change). And comparing the power variation delta PN of the photovoltaic module PVN with the voltage variation delta VN to obtain a data source delta PN/delta VN. By non-periodic, it is meant that the power and voltage variations of each photovoltaic module can be intermittently collected. For example, a first voltage value V of the nth photovoltaic module PVN is detected at some arbitrary random first TIME1_N-1 and a first output power value P_N-1 and collection of a second voltage value V of the nth photovoltaic module PVN at TIME2_N-2 and a second output power value P_N-2. After a certain unconfirmed TIME period of a certain TIME length after the previous data acquisition, the first voltage value V of the Nth photovoltaic module PVN is acquired at a certain random TIME TIME3_N-1 and a first output power value P_N-1 and collection of a second voltage value V of the nth photovoltaic module PVN at TIME4_N-2 and a second output power value P_N-2. It can be seen that the power variation and the voltage variation of the previous data collection occur between the first TIME1 and the second TIME2, and the power variation and the voltage variation of the subsequent data collection occur between the first TIME3 and the second TIME 4. And a period of unconfirmed TIME, namely, the power variation and the voltage variation of the photovoltaic module are intermittently acquired, which is substantially separated between the second TIME of the previous data acquisition TIME2 and the first TIME of the next data acquisition TIME 3.

Referring to fig. 5, in an alternative example, the power variation amount and the voltage variation amount of each photovoltaic module are calculated once every set period of time. For example, the power variation and the voltage variation of the pv module PVN are calculated once every such set TIME period or TIME width between the first TIME1 and the second TIME2, and the power variation and the voltage variation of the pv module PVN in the TIME period are Δ PN and Δ VN, respectively, so as to compare the power variation Δ PN of the pv module with the voltage variation Δ VN to obtain the data source Δ PN/Δ VN. Note that the TIME width of the set TIME period is adjustable, that is, the TIME width between the TIME1 at the first TIME and the TIME2 at the second TIME is adjustable, and different photovoltaic power plants or photovoltaic modules of different models can be designed with different TIME widths.

Referring to fig. 1, taking N photovoltaic modules PV1-PVN of a string as an example, it is assumed that a data collecting terminal or data collecting device is disposed at an inverter or a dc bus or at any position, and correspondingly, a data transmitting terminal or data transmitting device is disposed at each photovoltaic module. The data transmission equipment equipped with the photovoltaic modules firstly collects parameters such as voltage values, current values or power values of each photovoltaic module, and then the data transmission equipment equipped with the photovoltaic modules transmits the parameters to the data collection equipment in a wired communication mode, a wireless communication mode, a power line carrier mode and the like. The data collection equipment integrates the data sources of all the photovoltaic modules into one data set, and selects abnormal data sources which do not accord with preset rules. The biggest current doubt is that the communication opportunities between the data sending devices and the data collecting devices respectively equipped with the photovoltaic modules PV1-PVN in the group string have a chronological relationship. For example, the data collecting device sends out a command to inform the data sending device equipped with the photovoltaic module PV1 to feed back the parameters of the photovoltaic module PV1 to the data collecting device, and at this time, the data sending device of the photovoltaic module PV1 collects the parameters to the data collecting device. The data collection device then informs the data transmission device equipped with the photovoltaic module PV2 to feed back the parameters of the photovoltaic module PV2 to the data collection device, and the data transmission device of the photovoltaic module PV2 collects the parameters to the data collection device. The data collecting device informs the data transmitting device equipped with the photovoltaic module PV3 to feed back the parameters of the photovoltaic module PV3 to the data collecting device, and the data transmitting device of the photovoltaic module PV3 collects the parameters to the data collecting device. And by analogy, the data collection equipment enables the data sending equipment equipped by the PVN to feed back the parameters of the PVN to the data collection equipment, and at the moment, the data sending equipment of the PVN can acquire the parameters and send the parameters to the data collection equipment. The data collection device is equivalent to polling the data transmission devices respectively equipped with the photovoltaic modules PV1-PVN and then feeding back the parameter data to the data collection device through a series of data transmission devices. This doubt can lead to the fact that the parameters of the individual photovoltaic modules PV1-PVN in the string are not synchronous data: if the parameter collection time point of the photovoltaic module PV1 is earlier than that of the photovoltaic module PV2, a time difference exists; for example, the parameter collection time point of the photovoltaic module PV2 is earlier than that of the photovoltaic module PV3, and a time difference exists.

Referring to fig. 1, the output voltage, the output power and the current value of the photovoltaic module are changed at all times in the external changing environments such as the external real-time temperature, the irradiation intensity and the shielding condition. If the parameters of the individual PV modules PV1-PVN are not synchronous data then the individual members or objects of the final data set, i.e. the data sources, are not so synchronous data that the analysis results of the data set will have errors that are difficult to overcome due to the time differences between the individual data sources. After all, the parameters of the same photovoltaic module at different time points are different, and the parameters of voltage, power, current and the like are changed at any moment. Solution as context: the method comprises the steps of collecting a first voltage value and a first output power value of each photovoltaic module at a first moment, and collecting a second voltage value and a second output power value of each photovoltaic module at a second moment. And ensuring that all the photovoltaic modules synchronously acquire respective first voltage values and first output power values at a first moment and synchronously acquire respective second voltage values and second output power values at a second moment, so that the data sources of all the photovoltaic modules are acquired at the same time interval to form the data set in a data snapshot format. A data Snapshot (Database Snapshot) means that the data sets are of the same time period.

Referring to fig. 7, a first current value and a first output power value of each photovoltaic module are collected at a first TIME1 and a second current value and a second output power value of each photovoltaic module are collected at a second TIME 1. The description is temporarily illustrated by taking as an example the totaling N photovoltaic modules PV1-PVN among a single string: the first current value I of the first photovoltaic module PV1 is recorded at a first TIME point 1₁-1 and a first output power value P₁-1 and a second TIME instant TIME2 of acquisition of a second current value I of the first photovoltaic module PV1₁-2 and a second output power value P₁-2. The first TIME point 1 detects a first current value I of a second photovoltaic module PV2₂-1 and a first output power value P₂-1 and secondThe TIME2 detects a second current value I of a second photovoltaic module PV2₂-2 and a second output power value P₂-2. The first TIME TIME1 collects a first current value I of a third photovoltaic module PV3₃-1 and a first output power value P₃-1 and a second TIME instant TIME2 of acquiring a second current value I of a third photovoltaic module PV3₃-2 and a second output power value P₃-2. In the same way, the first current value I of the nth photovoltaic module PVN can be detected at the first TIME instant TIME1 in the string, and so on_N-1 and a first output power value P_N-1 and a second TIME instant TIME2 for acquiring a second current value I of the Nth photovoltaic module PVN_N-2 and a second output power value P_N-2. It has been clarified that the output power of each photovoltaic module is equal to the product of the current and the voltage, and the current of the photovoltaic module can be easily collected by the current detection module of the prior art, the voltage of the photovoltaic module can be easily collected by the voltage detection module of the prior art, and the current value and the output power value of the photovoltaic module can be collected by the prior art.

Referring to fig. 7, a first data source is obtained by comparing the power difference between the first and second output power values of the first photovoltaic module to the current difference between the first and second current values. For example, the power difference (i.e., P power difference) between the first and second output power values of the first photovoltaic module PV1₁-2 minus P₁-1) and the current difference between the first and second current values (current difference, i.e. I)₁-2 minus I₁-1) comparing, resulting in data source Δ P1/Δ I1. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P1 ═ P₁-2 minus P₁-1。ΔI1＝I₁-2 minus I₁-1。

Referring to fig. 7, a first data source is obtained by comparing the power difference between the first and second output power values of the second photovoltaic module to the current difference between the first and second current values. For example, the power difference (i.e., P power difference) between the first and second output power values of the second photovoltaic module PV2₂-2 minus P₂-1) and the current difference between the first and second current values (current difference, i.e. I)₂-2Subtract I₂-1) comparing, resulting in data source Δ P2/Δ I2. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P2 ═ P₂-2 minus P₂-1。ΔI2＝I₂-2 minus I₂-1。

Referring to fig. 7, a first data source is obtained by comparing the power difference between the first and second output power values of the third photovoltaic module to the current difference between the first and second current values. For example, the power difference (i.e., P power difference) between the first and second output power values of the third photovoltaic module PV3₃-2 minus P₃-1) and the current difference between the first and second current values (current difference, i.e. I)₃-2 minus I₃-1) comparing, resulting in data source Δ P3/Δ I3. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ P3 ═ P₃-2 minus P₃-1。ΔI3＝I₃-2 minus I₃-1。

Referring to fig. 7, the power difference between the first and second output power values of the nth photovoltaic module is compared to the current difference between the first and second current values to obtain a first data source. For example, the power difference (i.e., P) between the first and second output power values of the Nth PV module PVN_N-2 minus P_N-1) and the current difference between the first and second current values (current difference, i.e. I)_N-2 minus I_N-1) comparing, obtaining a data source Δ PN/Δ IN. The data source is negative, allowing the absolute value of the ratio to be taken as the data source. Δ PN ═ P_N-2 minus P_N-1。ΔIN＝I_N-2 minus I_N-1。

Referring to fig. 7, taking N photovoltaic modules PV1-PVN of a single string as an example, wherein data sources of all the photovoltaic modules are collected into one data set { Δ P1/Δ I1, Δ P2/Δ I2, Δ P3/Δ I3 … … Δ PN/Δ IN }, it is required to select an abnormal data source which does not meet a preset rule from the data sets { Δ P1/Δ I1, Δ P2/Δ I2, Δ P3/Δ I3 … … Δ PN/Δ IN }, and determine that the photovoltaic module whose data source has an abnormal power generation amount is IN an abnormal state.

Referring to fig. 7, the data sources of the N PV modules PV1-PVN may be grouped together in a single string as a data source for big data analysis, or the data sources of each of the different strings may be grouped together as a larger data set for big data analysis. For example, the data sets Δ P1/Δ I1- Δ PN/Δ IN of the first group of strings may be considered as the first data source set and the data sets Δ P1/Δ I1- Δ PN/Δ IN of the second group of strings may be considered as the second data source set, further, even the data sets Δ P1/Δ I1- Δ PN/Δ IN of the third group of strings may be considered as the third data source set and the data sets Δ P1/Δ I1- Δ PN/Δ IN of the fourth group of strings may be considered as the fourth data source set, if there are more group strings that can also provide data source sets of other group strings, the data sets Δ P1/Δ I1- Δ PN/Δ IN of the kth group string may be considered as the kth data source set and the positive integer K is greater than 1. According to the analysis principle: and a first data source set of the first group of strings, a second data source set of the second group of strings, a third data source set of the third group of strings, a fourth data source set of the fourth group of strings, … … and so on until a Kth data source set of the Kth group of strings and the like are collected to be used as a larger data set, and abnormal data sources which do not accord with preset rules need to be selected from the larger data set, so that the photovoltaic modules with abnormal data sources are judged to be in the state of abnormal power generation amount. The larger data set in this example is based on K x N data, while the single string is based on N data, in fact the more samples the more accurate results. Note that the sets of strings from which the dataset samples are taken may or may not have a parallel connection relationship between them.

Referring to fig. 7, a first current value and a first output power value of each photovoltaic module are collected at a first time, and a second current value and a second output power value of each photovoltaic module are collected at a second time. And all the photovoltaic modules synchronously acquire respective first current values and first output power values at a first moment, and all the photovoltaic modules synchronously acquire respective second current values and second output power values at a second moment, so that the data sources of all the photovoltaic modules are acquired at the same time interval to form the data set in a data snapshot format. A data Snapshot (Database Snapshot) means that the data sets are of the same time period.

Referring to fig. 5, note that the embodiment of fig. 1 and 7 are combined. Some string sets ST with faults are screened out from a plurality of string sets ST connected in parallel in the following way: the data sources of all the photovoltaic modules PV1-PVN in each string ST are added to obtain a power point characteristic value object, i.e. Δ P1/Δ V1+ Δ P2/Δ V2+ … … Δ PN/Δ VN, which is regarded as the so-called power point characteristic value object obtained by calculation. The first group of strings has a first power point eigenvalue object, the second group of strings has a second power point eigenvalue object, the third group of strings has a third power point eigenvalue object, the fourth group of strings has a fourth power point eigenvalue object, if there are more groups of strings, the eigenvalue objects of other groups of strings can be provided, for example, the power point eigenvalue object of the Kth group of strings is regarded as the Kth power point eigenvalue object. According to the principle of analysis: the first power point feature object of the first group string, the second power point feature object of the second group string, the third power point feature object of the third group string and the fourth power point feature object of the fourth group string, … …, and so on, until the K-th power point feature object of the K-th group string, etc., is collected as a data set or an object set, it is necessary to select an abnormal power point feature object that is inconsistent with the remaining other power point feature objects from the object set. The detection method of the power point feature value object with the same abnormality may be one of an abnormality point detection method based on statistics, an abnormality point detection method based on distances, an abnormality point detection method based on deviations, and an abnormality point detection method based on densities. Those group strings that can roughly be judged to have an abnormality (outlier) in the power point characteristic value object are in a failure state.

Referring to fig. 7, note that the embodiment of fig. 1 and 5 are combined. Some string sets ST with faults are screened out from a plurality of string sets ST connected in parallel in the following way: the data sources of all the photovoltaic modules PV1-PVN IN each string ST are added to obtain a power point characteristic value object, i.e. Δ P1/Δ I1+ Δ P2/Δ I2+ … … Δ PN/Δ IN is regarded as a so-called power point characteristic value object obtained by calculation. The first group of strings has a first power point eigenvalue object, the second group of strings has a second power point eigenvalue object, the third group of strings has a third power point eigenvalue object, the fourth group of strings has a fourth power point eigenvalue object, if there are more groups of strings, the eigenvalue objects of other groups of strings can be provided, for example, the power point eigenvalue object of the Kth group of strings is regarded as the Kth power point eigenvalue object. According to the principle of analysis: the first power point feature object of the first group string, the second power point feature object of the second group string, the third power point feature object of the third group string and the fourth power point feature object of the fourth group string, … …, and so on, until the K-th power point feature object of the K-th group string, etc., is collected as a data set or an object set, it is necessary to select an abnormal power point feature object that is inconsistent with the remaining other power point feature objects from the object set. The detection method of the power point feature value object with the same abnormality may be one of an abnormality point detection method based on statistics, an abnormality point detection method based on distances, an abnormality point detection method based on deviations, and an abnormality point detection method based on densities. Those group strings that can roughly be judged to have an abnormality (outlier) in the power point characteristic value object are in a failure state.

Referring to fig. 1, the main purpose of the MPPT function of pv inverter 200 is to set the bus voltage and bus current at the maximum power point, but it is essentially unknown whether or not pv inverter 200 is operating or operating at the maximum power point. An alternative example is how to determine whether the inverter is operating at the maximum power point as described below. The data sources for all photovoltaic modules under a single inverter 200 are summed to a sum. The data sources of the first group of strings are added to obtain a first value Δ P1/Δ V1+ Δ P2/Δ V2+ … … Δ PN/Δ VN, the data sources of the second group of strings are added to obtain a first value Δ P1/Δ V1+ Δ P2/Δ V2+ … … Δ PN/Δ VN, the first value Δ P1/Δ V1+ Δ P2/Δ V2+ … … Δ PN/Δ VN, the second value Δ P is added to obtain a second value Δ P1/Δ V1+ Δ P2/Δ V2+ … … Δ PN/Δ VN, the third value Δ P1/Δ V1+ Δ P2/Δ V2+ Δ V8253 Δ PN/Δ VN, the fourth value Δ VN is added to obtain a corresponding value Δ P1/Δ V1+ Δ P2/Δ V2+ Δ PN/Δ VN, and the corresponding value K is recorded by the data sources of the K group of the fourth string. The sum of the first value plus the second value plus the third value, … … until the sum to the Kth value is the sum of the data sources of all the photovoltaic modules under a single inverter, and the inverter is judged not to work at the maximum power point if the sum is not within the preset target value range by comparing the sum with the preset target value range.

Referring to fig. 1, another example of summing the values obtained by summing all the pv module data sources under inverter 200. The data sources of the first group of strings are added to obtain Δ P1/Δ I1+ Δ P2/Δ I2+ … … Δ PN/Δ IN, namely a first numerical value, the data sources of the second group of strings are added to obtain Δ P1/Δ I1+ Δ P2/Δ I2+ … … Δ PN/Δ IN, namely a second numerical value, the data sources of the third group of strings are added to obtain Δ P1/Δ I1+ Δ P2/Δ I2+ … … Δ PN/Δ IN, namely a third numerical value, the data sources of the fourth group of strings are added to obtain Δ P1/Δ I1+ Δ P2/Δ I2+ Δ PN/Δ IN, namely a fourth numerical value, and so on until the data sources of the K-th group of strings are added to obtain Δ P1/Δ I1+ Δ P2/Δ I2+ Δ I … … Δ PN/Δ IN, which is marked as a corresponding K-th numerical value. The sum of the first value plus the second value plus the third value, … … until the sum to the Kth value is the sum of the data sources of all the photovoltaic modules under a single inverter, and the inverter is judged not to work at the maximum power point if the sum is not within the preset target value range by comparing the sum with the preset target value range.

While the present invention has been described with reference to the preferred embodiments and illustrative embodiments, it is to be understood that the invention as described is not limited to the disclosed embodiments. Various alterations and modifications will no doubt become apparent to those skilled in the art after having read the above description. It is therefore intended that the appended claims be interpreted as covering all alterations and modifications as fall within the true spirit and scope of the invention. Any and all equivalent ranges and contents within the scope of the claims should be considered to be within the intent and scope of the present invention.

Claims

1. A method of diagnosing anomalies in electrical power production in a photovoltaic system, wherein a string providing direct current comprises a plurality of photovoltaic modules connected in series, the method comprising:

the data sources of all the photovoltaic modules are integrated into a data set, and abnormal data sources which do not accord with preset rules are selected from the data set, so that the photovoltaic modules with abnormal data sources are judged to be in a state of abnormal power generation amount;

providing direct current to an inverter from a plurality of strings connected in parallel, the inverter performing an inverse conversion of direct current to alternating current, while the inverter also performs maximum power tracking to operate the strings at a maximum power point;

the mode of screening the group string with faults is as follows:

2. The method of claim 1, wherein:

3. The method of claim 1, wherein:

the state of abnormality in power generation amount includes:

4. The method of claim 1, wherein:

the mode for judging the fault rate between different groups of strings is as follows:

5. The method of claim 1, wherein:

the method for determining whether the inverter operates at the maximum power point comprises the following steps:

6. The method of claim 1, wherein:

establishing normal distribution aiming at all data sources by utilizing the data set;

7. A method of diagnosing anomalies in electrical power production in a photovoltaic system, wherein a string providing direct current comprises a plurality of photovoltaic modules connected in series, the method comprising:

the data sources of all the photovoltaic modules are integrated into a data set, and abnormal data sources inconsistent with the rest of other data sources are selected from the data set, so that the photovoltaic modules with abnormal data sources are judged to be in a state of abnormal power generation amount;

the mode of screening the group string with faults is as follows:

8. The method of claim 7, wherein:

the method of selecting an abnormal data source from the data set is any one of an abnormal point detection method based on statistics, an abnormal point detection method based on distance, an abnormal point detection method based on deviation, and an abnormal point detection method based on density.

9. The method of claim 7, wherein:

the state of abnormality in power generation amount includes: the photovoltaic module with the data source with the abnormality does not work at the maximum power point.

10. A method for diagnosing an abnormal amount of power generation in a photovoltaic system, wherein a string for supplying direct current includes a plurality of photovoltaic modules connected in series, a plurality of strings for supplying direct current connected in parallel provides an inverter, the inverter performs an inverse conversion of direct current to alternating current, and the inverter further performs maximum power tracking to operate the strings at a maximum power point, the method comprising:

thereby judging that the photovoltaic module with the abnormal data source is in the state of abnormal power generation amount;

the mode of screening the group string with faults is as follows:

11. A method of diagnosing anomalies in electrical power production in a photovoltaic system, wherein a string providing direct current comprises a plurality of photovoltaic modules connected in series, the method comprising:

the mode of screening the group string with faults is as follows:

12. The method of claim 11, wherein:

the state of abnormality in power generation amount includes:

13. The method of claim 11, wherein: