Photovoltaic system monitoring and maintenance method, device and equipment and storage medium

Abstract

The invention discloses a photovoltaic system monitoring and maintenance method, device and equipment and a storage medium, and relates to the technical field of system monitoring, and the photovoltaic system monitoring and maintenance method comprises the specific steps: 1, collecting full-link operation data and environment data in real time through special sensors disposed at all core nodes of a photovoltaic system, 2, dividing a data monitoring time period according to the intraday change of the illumination intensity, constructing a reference correlation model of the illumination intensity and the component power based on historical data in a normal state, reducing the influence of environmental fluctuation, and obtaining a standard data set; and step 3, extracting core features from three dimensions of dynamic association, equipment collaboration and cumulative effect based on the reference model, dividing different feature state intervals, setting a severity index, and realizing fault positioning and hierarchical maintenance based on the severity index, equipment ID and space mapping.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of system monitoring, and particularly relates to a photovoltaic system monitoring and maintenance method, device, equipment and storage medium. BACKGROUND

[0002] With the rapid development of global new energy industry, as the core carrier of clean energy utilization, the installed capacity and application scale of photovoltaic systems continue to expand. However, photovoltaic systems are long-term operated in outdoor open environment, and are easily affected by dynamic fluctuations of natural factors such as light intensity, environmental temperature, wind speed, etc. At the same time, equipment problems such as component aging, inverter IGBT performance degradation, and storage battery cell voltage imbalance, will directly lead to the decline of system power generation efficiency and the rise of operation and maintenance cost, and even cause safety risks such as direct current arc and equipment burning. Therefore, realizing real-time monitoring, accurate fault identification and efficient maintenance of photovoltaic systems has become the core demand to ensure their stable operation in the whole life cycle and to improve power generation efficiency.

[0003] The current photovoltaic system monitoring and maintenance technology still has many key defects, and it is difficult to meet the actual needs of efficient operation and maintenance of large-scale photovoltaic power stations. The existing monitoring method mainly adopts fixed time interval for data collection and analysis, and the single monitoring strategy is easy to misjudge the "normal power fluctuation caused by environmental factors" as equipment failure, or to cover up the hidden fault signals such as "component hidden crack and virtual connection" due to environmental interference, which greatly reduces the accuracy of abnormal identification. The existing technology mainly judges the equipment running state from a single dimension, which is easy to miss the hidden faults such as component hidden crack and storage battery cell voltage imbalance.

[0004] Therefore, there is an urgent need for a photovoltaic system monitoring and maintenance method to solve the above problems. SUMMARY

[0005] The purpose of the present application is to provide a photovoltaic system monitoring and maintenance method, device, equipment and storage medium, which solves the technical problem of single abnormal identification dimension in the prior art, ignoring the three key dimensions of dynamic correlation, equipment cooperation and cumulative effect.

[0006] In order to achieve the above purpose, the application adopts the following technical scheme:

[0007] The photovoltaic system monitoring and maintenance method, device, equipment and storage medium comprise:

[0008] Step one, through the special sensors deployed at each core node of the photovoltaic system, real-time collection of full-link operation data and environmental data, and through data cleaning and standardization processing, a standardized data set is obtained;

[0009] The photovoltaic module parameters include module output power, module backboard temperature, module open-circuit voltage and short-circuit current, the inverter parameters include inverter input voltage and output current, inverter conversion efficiency, inverter heat dissipation temperature, the energy storage device parameters include energy storage battery charging and discharging current, energy storage battery SOC value, energy storage battery cell voltage balance degree, and the environmental parameters include real-time light intensity, environmental temperature and wind speed;

[0010] Step two, the data monitoring time period is divided according to the intra-day change of light intensity, and a reference correlation model of light intensity and module power is constructed based on historical data under normal state, so as to reduce the influence of environmental fluctuations.

[0011] Step three, based on the reference model, core features are extracted from three dimensions of dynamic correlation, device cooperation and cumulative effect, different feature state intervals are divided, a severity index is set, and fault positioning and hierarchical maintenance are realized based on the severity index, device ID and space mapping.

[0012] Further, the data monitoring time period is divided according to the intra-day change of light intensity, and the specific method is:

[0013] According to the intra-day change rule of light intensity G, 1 day is divided into n monitoring time periods The light intensity monitoring range [a, b] is set, wherein a is the starting threshold value, equal to the minimum light intensity at which the photovoltaic module starts to output effective power, and below a, there is no need for monitoring, b is the termination threshold value, when the light intensity is greater than b, the light intensity enters the flat growth interval, the module power rising rate slows down, and the true values of a and b are determined according to historical data and actual situation.

[0014] Further, a reference correlation model of light intensity and module power is constructed based on historical data under normal state, and the specific method is:

[0015] The normal state judgment condition is set, the linear regression equation P=k×G+b of light intensity and module output power is determined, wherein k is the power-light coefficient, and b is the intercept, the light intensity and module output power data of continuous N1 days under normal state are combined to form a set D{normal}, the light intensity in D{normal} is grouped according to H interval, the average value of light intensity in each group is calculated And the average value of corresponding module output power The sum of squared errors The value of k is solved with the minimum sum of squared errors as the target, and the value of b is determined according to the solved value of k, so as to realize the construction of the reference correlation model.

[0016] Further, the normal state judgment condition is set, and the specific method is:

[0017] The normal state refers to data of a time period meeting three core conditions of no fault of the photovoltaic module device, no interference of the environment, and no abnormality in operation;

[0018] The no fault of the device refers to no any operation and maintenance fault repair record in the time period, and no abnormal alarm signal of the component-level monitor;

[0019] The no interference of the environment refers to that the light intensity G is in the effective power generation light range of the photovoltaic module, the light fluctuation amplitude is less than a preset fluctuation interference amplitude threshold, the environmental temperature and the wind speed are in the normal range, and the normal range is set by historical data and actual situation;

[0020] The no abnormality in operation refers to that the component output power in the time period has no power jump caused by intermittent fault, and has no short-term rapid attenuation.

[0021] Further, based on the benchmark model, core features are extracted from three dimensions of dynamic correlation, device cooperation and cumulative effect, and different feature state intervals are divided, and the specific method is as follows:

[0022] The output power prediction value of the component in the tn time period is determined through the benchmark correlation model, the actual power of the component is compared, and the formula is used to determine the mean value of the relative deviation of the predicted power;

[0023] Wherein, represents the mean value of the relative deviation of the predicted power in the tn time period, represents the actual power of the component at the i-th sampling point, represents the output power prediction value of the component at the i-th sampling point, and W represents the number of sampling points in the tn time period;

[0024] The light-power correlation state judgment standard is set, when A1 represents normal, when A1 <A2 represents observation, when represents abnormal;

[0025] The ratio of the standard deviation of the output power of all components in the string in the tn time period to the average power of the string is calculated, and is recorded as ;

[0026] The string power state judgment standard is set, when C1 represents normal, when C1 C2 represents observation, when represents abnormal;

[0027] The power attenuation rate based on the continuous n1 time periods is calculated by the second-order difference to calculate the attenuation speed change, which is recorded as ap;

[0028] The power attenuation state judgment standard is set, when E0 E1 represents normal, when E1 E2 represents to be observed, when E2 represents abnormal, A1, A2, C1, C2, E0, E1 and E2 are constant values, which are set according to historical data and actual situation.

[0029] The inverter state is used to determine the inverter abnormal condition, and the battery SOC state is used to determine the energy storage abnormal condition.

[0030] Further, the inverter state is used to determine the inverter abnormal condition, and the battery SOC state is used to determine the energy storage abnormal condition, and the specific method is:

[0031] The inverter heat dissipation temperature and the conversion efficiency covariance and standard deviation ratio, denoted as ;

[0032] The temperature-efficiency coupling state determination standard is set, when B2 B1 represents normal, when B1 <B0 represents to be observed, when represents abnormal, wherein B1, B2 and are constant values, which are set according to historical data and actual situation.

[0033] The SOC value is divided into N2 intervals, and the information entropy Hs of each interval probability is calculated.

[0034] The SOC fluctuation state determination standard is set, when <D1 represents normal, when D1 D2 represents to be observed, when represents abnormal, D1, are constant values, which are set according to historical data and actual situation.

[0035] Further, the severity index is set, and the specific method is:

[0036] The severity index Ss is defined as k1×Sr+k2×Sl, wherein Sr represents the abnormal state index, and Sl represents the daily power generation loss rate;

[0037] In a time period, when no component, inverter or battery appears abnormal state, Sr is equal to 0, the value of Sr is added by a1 for each component appearing abnormal state, the value of Sr is added by a2 for each inverter appearing abnormal state, and the value of Sr is added by a3 for each single battery appearing abnormal state, a1, a2 and a3 are set according to historical data and actual situation, k1 represents the abnormal state index weight, and k2 represents the daily power generation loss rate weight, which is set based on the historical fault handling records of the power station and actual situation.

[0038] The application also provides a photovoltaic system monitoring and maintenance device, specifically comprising:

[0039] a data acquisition and processing module, a time period division and model construction module, an abnormality identification and positioning module, and a hierarchical maintenance module;

[0040] The data acquisition and processing module is used to collect multi-dimensional parameters through a special sensor and perform abnormal value elimination, missing value filling, and normalization processing;

[0041] The time period division and model construction module is used to divide monitoring time periods according to illumination changes, and construct and verify an illumination intensity-component power benchmark correlation model based on historical normal data;

[0042] The abnormality identification and positioning module is used to extract core features, classify and judge abnormal types, and realize fault positioning through device ID and spatial mapping;

[0043] The hierarchical maintenance module is used to calculate a fault severity index, and generate and execute a hierarchical maintenance strategy.

[0044] The application also provides a photovoltaic system monitoring and maintenance device, comprising a processor, a memory, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the steps of the photovoltaic system monitoring and maintenance method are implemented.

[0045] The application also provides a computer readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the photovoltaic system monitoring and maintenance method are implemented; and the storage medium comprises at least one of a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk, or an optical disk.

[0046] As described above, due to the adoption of the above technical solutions, the application has the following beneficial effects:

[0047] 1. The application improves the accuracy and reliability of photovoltaic system abnormality identification in a systematic manner through time period monitoring, strict normal state data screening, benchmark correlation model construction, and error verification, reduces the interference of environmental fluctuations on power monitoring through segmented management, improves the sensitivity and accuracy of abnormality identification, and provides a reliable reference for abnormality identification through strict screening, and the benchmark model constructed can truly reflect the performance of the component in a normal state.

[0048] 2、The present application constructs the benchmark correlation model of light intensity and component power based on historical data under normal state, effectively excludes the influence of various interference factors on the model by screening data under the three core conditions of no failure of equipment, no interference of environment and no abnormal operation, makes the model more accurately reflect the real relationship between light intensity and component power, improves the reliability and accuracy of the model, constructs the benchmark correlation model with linear regression equation as the core, calculates the average value of light intensity by grouping through selecting historical operation data under normal state for continuous days, solves the power-light coefficient and intercept with the minimum sum of squared errors as the target, the scientific modeling method guarantees the rationality and effectiveness of the model, and the linear relationship between light intensity and component power can be well fitted.

[0049] 3、The present application realizes accurate positioning and differential maintenance of photovoltaic system failure by extracting core features from three dimensions of dynamic correlation, equipment cooperation and cumulative effect, realizes early detection, accurate positioning and differential processing of photovoltaic system failure by integrating abnormal state indicators and daily power generation loss rate to determine the severity index quantization of fault influence, and improves the operation and maintenance efficiency and system reliability. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0051] Figure 1 The present application photovoltaic system monitoring and maintenance method step diagram is shown;

[0052] Figure 2 The present application hierarchical maintenance strategy determination method step diagram is shown;

[0053] Figure 3 The present application photovoltaic system monitoring and maintenance device module diagram is shown. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0055] Embodiment one, as Figure 1The photovoltaic system monitoring and maintenance method, device, equipment and storage medium shown specifically comprises the following steps:

[0056] Step one, through the special sensors deployed in each core node of the photovoltaic system, real-time collection of full-link operation data and environmental data, and through data cleaning and standardization processing, a standardized data set is obtained;

[0057] Among them, the collection method of each parameter, sensor selection and deployment logic are as follows:

[0058] Photovoltaic module parameter collection, photovoltaic module parameters include module output power, module backboard temperature, module open circuit voltage and short circuit current;

[0059] Among them, a component-level power sensor is used, one is connected in series at the positive and negative output terminals of each component, the voltage and current output by the component are collected in real time, the instantaneous power is calculated, and the average value of the component output power in the time period is taken as the component output power value in the time period;

[0060] A patch type PT100 temperature sensor is used, which is pasted at the center position of the backboard of each component (avoiding the junction box area to avoid local high temperature interference), and through the temperature change characteristic of platinum resistance value, the temperature value is output in real time (the sampling frequency of this embodiment is 1 minute / time);

[0061] A component-level IV curve monitoring sensor is used, which is deployed at the inlet end of each component string, and the IV curve scanning of the component is triggered periodically (once every 30 minutes in this embodiment) to obtain the voltage in the open circuit state and the current in the short circuit state;

[0062] Inverter parameter collection, inverter parameters include inverter input voltage and output current, inverter conversion efficiency, inverter heat dissipation temperature;

[0063] Among them, a Hall voltage sensor and a Hall current sensor are used, which are respectively installed on the direct current input side of the inverter (one on each input side) and the alternating current output side (one on each output side), and the voltage and current signals are collected in real time to obtain the inverter input voltage and output current;

[0064] Based on the above input voltage, input current, output voltage and output current (the output voltage is provided by the grid side sensor), the inverter output AC power P1 and the inverter input DC power P2 are calculated, and the inverter conversion efficiency is obtained by calculating the ratio of P1 and P2;

[0065] A duct type NTC temperature sensor is used, which is installed at the outlet of the heat dissipation fan of the inverter, and the air temperature in the heat dissipation duct is collected in real time to obtain the inverter heat dissipation temperature.

[0066] Energy storage device parameter acquisition, energy storage device parameters include energy storage battery charge and discharge current, energy storage battery SOC value, energy storage battery single cell voltage balance degree;

[0067] Among them, high-precision shunt is adopted, which is connected in series in the charge and discharge main circuit of the energy storage battery pack. The voltage difference between the two ends of the shunt is collected, and the energy storage battery charge and discharge current is converted according to Ohm's law;

[0068] Based on the real-time acquisition data of energy storage battery charge and discharge current, the energy storage battery SOC value is calculated by the "ampere-hour integration method" built in BMS;

[0069] Multi-channel voltage acquisition is adopted, and each battery single cell is connected with one acquisition channel at the positive and negative electrodes. The voltage of each single cell is collected in real time, and the difference between the highest single cell voltage value and the lowest single cell voltage value is calculated to obtain the energy storage battery single cell voltage balance degree.

[0070] Environmental parameter acquisition, environmental parameters include real-time illumination intensity, environmental temperature and wind speed;

[0071] Among them, silicon-based irradiation sensor is adopted, which is deployed in the south unobstructed area of photovoltaic array (the height is consistent with the inclination angle of the component), and the solar irradiation is converted into current signal by silicon photocell, and then the illumination intensity G is converted;

[0072] Anti-radiation temperature sensor (with sunshade) is adopted, which is deployed at the same position as the irradiation sensor, and the real-time environmental air temperature is collected;

[0073] Cup-type wind speed sensor is adopted, which is deployed at the high point of photovoltaic power station (such as the top of monitoring tower), and the real-time environmental wind speed is collected.

[0074] Through 3σ criterion, abnormal values are eliminated, linear interpolation method is used to fill in the offline data of sensor in a short time (such as 10 minutes), all acquisition parameters are normalized to eliminate dimension difference, and standardized multi-dimensional parameter data set is obtained.

[0075] Step two, according to the division of data monitoring time period according to the daily change of illumination intensity, the reference correlation model of illumination intensity and component power is constructed based on the historical data under normal state, and the influence of environmental fluctuation is reduced;

[0076] According to the daily change rule of illumination intensity G, 1 day is divided into n monitoring time periods , the monitoring range of illumination intensity [a, b] is set, where a is the starting threshold value, which is equal to the minimum illumination intensity at which the photovoltaic component starts to output effective power. Below this value, the component power is close to 0, and there is no need for monitoring. B is the termination threshold value. When the illumination intensity is greater than b, the illumination intensity enters the flat growth interval, and the component power rises slowly.

[0077] This embodiment divides 1 day into 3 monitoring time periods (n=1, 2, 3), light rising segment ): 6:30-9:00 (summer) / 7:30-10:00 (winter), in this stage G rises from 50 W / m2 to above 800 W / m2, the module power rises rapidly, and it is necessary to monitor whether the power changes synchronously with G; light stable segment ): 9:00-16:00 (summer) / 10:00-15:00 (winter), in this stage G maintains at 800-1000 W / m2, the system runs stably, and it is the core time period for anomaly identification, and the power decay rate and efficiency fluctuation value can be accurately calculated; light falling segment ): 16:00-19:00 (summer) / 15:00-18:00 (winter), in this stage G falls from 800 W / m2 to below 50 W / m2, the module power slowly falls, and it is necessary to monitor whether the power falls synchronously with G to avoid misjudgment of failure due to power falling.

[0078] The standardized data set divided by time period (D={D1, D2,..., Dn}, Dn is a multi-dimensional parameter set of the nth time period) is determined.

[0079] Based on the preprocessed data set, the key operation characteristics of the photovoltaic system are extracted,

[0080] A reference correlation model of light intensity G and module output power in a normal state is established based on historical data;

[0081] The normal state refers to a time period that meets three core conditions of no failure of photovoltaic module equipment, no environmental interference, and no abnormal operation.

[0082] Further, no failure of equipment means that there is no any operation and maintenance failure repair record (such as shading cleaning and module replacement) in the time period, and the component-level monitor has no abnormal alarm signal.

[0083] No environmental interference means that the light intensity G is in the effective photovoltaic module power generation light range, the light fluctuation amplitude is less than a preset fluctuation interference amplitude threshold, and the environmental temperature and wind speed are within a normal range (the normal range is set by historical data and actual situation);

[0084] No abnormal operation means that the module output power has no power jump caused by intermittent failure and no short-term rapid decay in the time period.

[0085] Further, the reference correlation model of light intensity G and module output power takes a linear regression equation as the core, selects historical operation data in a normal state for continuous N1 days, and determines the frequency-synchronous light intensity and the module output power set D{normal}={( , ), , ), , }, wherein m represents the mth historical time period, a linear reference correlation model is constructed with the irradiance as the independent variable and the component output power , wherein k is the power-irradiance coefficient (reflecting the power output capability under unit irradiance), and b is the intercept (theoretically, the power loss when the irradiance is 0).

[0086] Group G in D{normal} according to the interval of H (such as H=50W / m²) (such as 200-250, 250-300,..., 950-1000W / m²), calculate the average value of irradiance in each group and the average value of the corresponding component output power ;

[0087] Let the average value data after grouping be , ), ( , ),..., ( , ) (M is the number of groups);

[0088] Take the minimum sum of squares of errors as the target, that is min, solve k, and the functional expression of k is , and then determine the value of b according to the solved k value, and the functional expression of b is , substitute the solved k and b into the linear equation to obtain the reference correlation model of the component under normal state.

[0089] Randomly select 20% of the data from D{normal} as the test set, and the remaining part as the training set. Calculate the output power corresponding to each irradiance in the test set using the model constructed by the training set, and record it as the output power prediction value. Calculate the prediction error rate ε by comparing the real output power value of the test set. When the ε of more than 90% of the data in the test set is less than 3%, it means that the reference correlation model training error is low, otherwise, reduce the reference correlation model training error by prolonging the data collection period or optimizing the screening standard of the normal state. 3% when more than 90% of the data in the test set is less than 3%, it means that the reference correlation model training error is low, otherwise, reduce the reference correlation model training error by prolonging the data collection period or optimizing the screening standard of the normal state.

[0090] Step three, based on the reference model, extract core features from three dimensions of dynamic correlation, device collaboration, and cumulative effect, divide different feature state intervals, set the severity index, and realize fault positioning and hierarchical maintenance based on the severity index, device ID, and space mapping;

[0091] Determine the abnormal condition of the component based on the component state;

[0092] Determine the predicted output power value of the component in the tn time period through the reference correlation model, compare the actual power of the component, and determine the average relative deviation of the predicted power. The specific calculation formula is:

[0093] ;

[0094] Where, represents the average relative deviation of the predicted power in the tn time period, represents the actual power of the component at the i-th sampling point, represents the predicted output power value of the component at the i-th sampling point, and W represents the number of sampling points in the tn time period;

[0095] Set the determination criteria for the light intensity-power correlation status:

[0096] When A1, it indicates normal. When A1 < < A2, it indicates to be observed. When , it indicates abnormal. Among them, A1 takes the allowable deviation of the rated power of the component (refer to the common requirements for photovoltaic component procurement. For example, in a certain procurement project, it is clearly stated that "the output power deviation of the component ≤ +3%"), and A2 takes the minimum average value of the power deviation triggered by hidden faults (such as hidden cracks and slight shading) in the historical fault data;

[0097] Calculate the ratio of the standard deviation of the output power of all components in the string to the average power of the string in the tn time period, denoted as ;

[0098] Set the determination criteria for the string power status:

[0099] When C1 , it indicates normal. When C1 C2, it indicates to be observed. When , it indicates abnormal. C1 takes the power consistency threshold of the newly commissioned component string (based on the factory test data of the same batch of components), and C2 refers to the critical value of the damage to the string consistency caused by single-component faults (such as loose wiring and hidden cracks) in the operation and maintenance historical data;

[0100] Calculate the change in the attenuation speed through the second-order difference based on the power attenuation rate in n1 consecutive time periods, denoted as ap;

[0101] Set the determination criteria for the power attenuation status:

[0102] When E0 E1, it indicates normal. When E1 E2, it indicates to be observed. When E2 , it indicates abnormal. E0, E1, and E2 are set according to historical data and actual situations;

[0103] Determine inverter abnormal condition based on inverter state;

[0104] Calculate inverter heat dissipation temperature Covariance of conversion efficiency , denoted as ;

[0105] Set temperature-efficiency coupling state determination criteria:

[0106] B2 B1 represents normal, when B1 B0 represents observation, when B1, B2 take the temperature-efficiency negative correlation interval of normal inverter, and Take the critical value when IGBT aging or heat dissipation failure reduces the temperature-efficiency negative correlation;

[0107] Determine energy storage abnormal condition based on battery SOC state;

[0108] Divide the SOC value into N2 intervals (such as 0-5%, 5%-10%, …, 95%-100%), and calculate the information entropy Hs of each interval probability;

[0109] Set SOC fluctuation state determination criteria:

[0110] When D1 represents normal, when D1 D2 represents observation, when D1, D2 take the upper limit of SOC fluctuation entropy during normal charging and discharging, Take the critical value when battery aging or single cell short circuit leads to fluctuation disorder;

[0111] Based on the unique ID of the device and the space mapping to realize fault positioning;

[0112] Define severity index Ss=k1×Sr+k2×Sl, where Sr represents abnormal state index, Sl represents daily power generation loss rate (obtained by comparing actual power generation during abnormal state period with theoretical power generation during normal period), k1 represents abnormal state index weight, k2 represents daily power generation loss rate weight, based on historical fault handling records and actual situation of power station;

[0113] In a time period, when no component, inverter or battery appears abnormal state, Sr is equal to 0, the value of Sr will increase by a1 for each component that appears abnormal state, the value of Sr will increase by a2 for each inverter that appears abnormal state, and the value of Sr will increase by a3 for each single battery that appears abnormal state, a1, a2 and a3 are obtained according to historical data and actual situation;

[0114] Set hierarchical maintenance strategy, emergency failure (S1 Ss): Remote disconnect power supply of failed equipment (such as component group string DC switch, inverter shutdown), push location report + tool list + GIS navigation;

[0115] General failure (S2 Ss<S1): Reserve a maintenance window (such as "next day 9:00-11:00") for the maintenance personnel with adaptation experience (such as technicians familiar with specific model inverters), and continuously monitor the failure state before maintenance;

[0116] Minor failure (Ss<S2): Include weekly inspection, generate targeted prompts (such as "use a soft brush to clean the component surface dust, avoid scratching the glass cover plate"), and upgrade to general failure if it does not recover for 2 weeks.

[0117] Embodiment two, as Figure Three The photovoltaic system monitoring and maintenance device comprises:

[0118] A data acquisition and processing module is configured to acquire real-time full-link operation data and environmental data through special sensors deployed at core nodes of the photovoltaic system, and perform outlier rejection, missing value filling and normalization processing;

[0119] A time period division and model construction module is configured to divide monitoring time periods according to light changes, and construct and verify a light intensity-component power benchmark correlation model based on historical normal data;

[0120] An abnormality identification and positioning module is configured to extract core features, classify and determine abnormal types, and realize fault positioning through device ID and spatial mapping;

[0121] A hierarchical maintenance module is configured to calculate a fault severity index, and generate and execute a hierarchical maintenance strategy.

[0122] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can make equivalent replacements or changes to the technical solution and the inventive concept of the present application within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application.

[0123] The preferred embodiments of the application disclosed above are only used to help explain the application. The preferred embodiments do not describe all the details, nor limit the application to the specific embodiments. Obviously, many modifications and changes can be made according to the content of the present application. The present application selects and describes these embodiments in order to better explain the principles and practical applications of the present application, so that those skilled in the art can well understand and utilize the present application. The present application is limited by the claims and their entire scope and equivalents.

Claims

1. A method for monitoring and maintaining a photovoltaic system, characterized in that, include: Step 1: Collect real-time operational and environmental data across the entire photovoltaic system using dedicated sensors deployed at each core node. Then, through data cleaning and standardization, obtain a standardized dataset. The photovoltaic module parameters include module output power, module backsheet temperature, module open-circuit voltage and short-circuit current; the inverter parameters include inverter input voltage and output current, inverter conversion efficiency and inverter heat dissipation temperature; the energy storage device parameters include energy storage battery charging and discharging current, energy storage battery SOC value and energy storage battery single cell voltage uniformity; and the environmental parameters include real-time light intensity, ambient temperature and wind speed. Step 2: Divide the data monitoring time period according to the intraday variation of light intensity, and build a benchmark correlation model between light intensity and component power based on historical data under normal conditions to reduce the impact of environmental fluctuations; Step 3: Based on the benchmark model, extract core features from three dimensions: dynamic correlation, equipment collaboration, and cumulative effect. Divide different feature state intervals, set a severity index, and realize fault location and graded maintenance based on the severity index, equipment ID and spatial mapping.

2. The photovoltaic system monitoring and maintenance method according to claim 1, characterized in that, The data monitoring period is divided according to the intraday variation of light intensity. The specific method is as follows: Based on the diurnal variation pattern of light intensity G, one day is divided into n monitoring time periods. Set the light intensity monitoring range [a, b], where a is the starting threshold, which is equal to the minimum light intensity at which the photovoltaic module begins to output effective power. There is no need to monitor when the light intensity is below a. b is the termination threshold. When the light intensity is greater than b, the light intensity enters a gradual growth range and the rate of increase in module power slows down. The actual values ​​of a and b are determined based on historical data and actual conditions.

3. The photovoltaic system monitoring and maintenance method according to claim 1, characterized in that, A baseline correlation model between light intensity and component power is constructed based on historical data under normal conditions. The specific method is as follows: Set the criteria for determining normal state, and determine the linear regression equation P=k×G+b for light intensity and component output power, where k is the power minus the illuminance coefficient and b is the intercept. Collect the light intensity and component output power data under normal state for N1 consecutive days to form a set D{normal}. Group the light intensity in D{normal} according to the interval H, and calculate the average light intensity within each group. Average value of the output power of the corresponding components Sum of squared errors The goal is to minimize the value of k, and then determine the value of b based on the obtained value of k, thereby constructing the baseline correlation model.

4. The photovoltaic system monitoring and maintenance method according to claim 3, characterized in that, The method for setting the conditions for determining the normal state is as follows: Normal state refers to the data within a time period that meets the three core conditions of no faults in photovoltaic module equipment, no environmental interference, and no abnormal operation; Equipment without faults means that there are no maintenance fault reports or repair records during this period, and the component-level monitors have no abnormal alarm signals. "No environmental interference" means that the light intensity G is within the effective power generation range of the photovoltaic module during this period, the light fluctuation amplitude is less than the preset fluctuation interference amplitude threshold, and the ambient temperature and wind speed are within the normal range. The specific value range of the normal range is set by historical data and actual conditions. "No abnormal operation" means that during this period, the component output power did not experience intermittent power fluctuations caused by faults, and there was no short-term rapid attenuation.

5. The photovoltaic system monitoring and maintenance method according to claim 1, characterized in that, Based on the baseline model, core features are extracted from three dimensions: dynamic correlation, equipment collaboration, and cumulative effect. Different feature state intervals are then defined. The specific method is as follows: The predicted output power of the component during the time period tn is determined using a benchmark correlation model, compared with the actual power of the component, and then calculated using the formula. Determine the mean of the relative deviation of the predicted power; in, This represents the mean relative deviation of predicted power over the time period tn. This represents the actual power of the component at the i-th sampling point. Let W represent the predicted output power of the component at the i-th sampling point, and let W represent the number of sampling points in the time period tn. Set the determination criteria for the light intensity-power correlation state. When A1, it indicates normal. When A1 < < A2, it indicates to be observed. When , it indicates abnormal; The ratio of the standard deviation of the output power of all components in the string during the time period tn to the average power of the string is denoted as . ; Set the string power status judgment criteria, when C1 When C1 is normal, it indicates that the time is normal. C2 indicates that it is to be observed. Time indicates an anomaly; The change in attenuation rate is calculated by second-order difference based on the power attenuation rate over n1 consecutive time periods, and is denoted as ap. Set the power attenuation status judgment criteria, when E0 E1 indicates normal operation. E2 indicates that it is to be observed. Indicates an anomaly. A1, A2, C1, C2, E0, E1, and E2 are all constant values, set based on historical data and actual conditions. Inverter-related anomalies are determined based on inverter status, and energy storage-related anomalies are determined based on battery SOC status.

6. The photovoltaic system monitoring and maintenance method according to claim 5, characterized in that, Inverter-related anomalies are determined based on inverter status, and energy storage-related anomalies are determined based on battery SOC status. The specific methods are as follows: Calculate the inverter's heat dissipation temperature With conversion efficiency The ratio of covariance to standard deviation is denoted as . ; Set the temperature-efficiency coupling state determination criteria. When B2 < B1, it indicates normal. When B1 < < B0, it indicates pending observation. When it indicates abnormal, where B1, B2 and are all constant values, which are set according to historical data and actual situations; Divide the SOC value into N2 intervals and calculate the information entropy Hs of the probability of each interval. Set the judgment criteria for the SOC fluctuation state. When < D1 indicates normal. When D1 ≥ D2 indicates pending observation. When ≥, it indicates abnormal. D1 and are both constant values, which are set according to historical data and actual situations.

7. The photovoltaic system monitoring and maintenance method according to claim 1, characterized in that, To set the severity index, the specific method is as follows: The severity index Ss is defined as k1×Sr+k2×Sl, where Sr represents the abnormal state index and Sl represents the daily power generation loss rate. Within a given time period, when no components, inverters, or batteries exhibit abnormal conditions, Sr equals 0. For each component exhibiting an abnormal condition, the Sr value is incremented by a1; for each inverter exhibiting an abnormal condition, the Sr value is incremented by a2; and for each individual battery exhibiting an abnormal condition, the Sr value is incremented by a3. a1, a2, and a3 are set based on historical data and actual conditions. k1 represents the weight of the abnormal condition index, and k2 represents the weight of the daily power generation loss rate, which is set based on the power station's historical fault handling records and actual conditions.

8. A photovoltaic system monitoring and maintenance device, characterized in that, The photovoltaic system monitoring and maintenance method according to any one of claims 1-7 specifically includes: Data acquisition and processing module, time period segmentation and model building module, anomaly identification and location module, and hierarchical maintenance module; The data acquisition and processing module is used to collect multi-dimensional parameters through dedicated sensors and perform outlier removal, missing value imputation and normalization. The time period segmentation and model building module is used to segment monitoring time periods according to changes in illumination, and to build and verify a baseline correlation model of illumination intensity and component power based on historical normal data. The anomaly identification and localization module is used to extract core features, classify and determine the anomaly type, and realize fault localization through device ID and spatial mapping; The graded maintenance module is used to calculate the fault severity index and generate and execute graded maintenance strategies.

9. Photovoltaic system monitoring and maintenance equipment, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory. When the computer program is executed by the processor, it implements the steps of the photovoltaic system monitoring and maintenance method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the photovoltaic system monitoring and maintenance method according to any one of claims 1-7; the storage medium includes at least one of a USB flash drive, a portable hard drive, a read-only memory, a random access memory, a magnetic disk, or an optical disk.

Images