An intelligent optimization system and method for a photovoltaic power station based on an internet of things

Abstract

The application discloses a kind of photovoltaic power station intelligent optimization system and method based on Internet of Things, it is related to photovoltaic power station intelligent optimization technical field, the method includes the following steps: obtaining the historical operation data of target photovoltaic power station to be optimized power station unit, constructs each power station unit dataset;The health degree stress index of each statistical period is calculated, compared with preset threshold, the period that exceeds threshold is screened out, forms the historical health degree abnormal period set of target power station unit, with power station real-time operation data, with the shortest estimated repair time as target generates candidate optimization strategy, calculate instruction estimated completion time and judge whether it is in corresponding abnormal period, if it is in then the average value of historical health degree stress index of the period is revised estimated length, traverse each strategy, the number of actions after time consumption that exceeds original estimated length is counted, the matching value of strategy and power station current state is calculated, the optimal strategy is recommended after sorting according to matching value, the application realizes photovoltaic power station intelligent optimization.

Description

Technical Field

[0001] This invention relates to the field of intelligent optimization technology for photovoltaic power plants, specifically an intelligent optimization system and method for photovoltaic power plants based on the Internet of Things. Background Technology

[0002] With the large-scale development of the photovoltaic industry, the long-term and efficient operation of photovoltaic power plants has become a core requirement of the industry. The health status management and operation and maintenance strategy optimization of key power plant units such as strings and inverters directly affect the power generation efficiency and equipment lifespan of the power plant, and are the core links to ensure the stable income of the power plant.

[0003] However, traditional photovoltaic power plant operation and maintenance methods often face the following problems when dealing with complex operational data processing, equipment health assessment, and operation and maintenance strategy optimization: First, operation and maintenance efficiency is low and response is slow. Traditional operation and maintenance relies heavily on manual inspections or simple data monitoring. Faced with massive historical and real-time operational data of the power plant, manual processing is time-consuming and labor-intensive, making it difficult to quickly extract key information, and there is a delay in identifying abnormal equipment conditions. Second, there is a lack of health assessment and personalized strategy generation capabilities. Traditional methods mostly judge equipment status based on single operational indicators, without combining historical power generation, performance ratio, and other multi-dimensional data to comprehensively calculate the health stress index. The formulation of operation and maintenance strategies is highly subjective and does not consider the impact of abnormal equipment periods on operation and maintenance duration, resulting in insufficient strategy targeting. In addition, operation and maintenance strategies lack dynamic correction and adaptability assessment. After traditional strategies are formulated, the estimated operation and maintenance duration is not dynamically adjusted according to the actual health status of the equipment, and no matching assessment mechanism between the strategy and the current overall status of the power plant is established, resulting in poor strategy execution and failure to maximize the improvement of power plant operation efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent optimization system and method for photovoltaic power plants based on the Internet of Things, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a smart optimization method for photovoltaic power plants based on the Internet of Things, the method comprising the following steps:

[0006] Obtain historical operating data of the target photovoltaic power plant's units to be optimized, and construct datasets for each power plant unit;

[0007] Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out the periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit;

[0008] Based on real-time operation data of photovoltaic power plants, candidate optimization strategies are generated with the goal of minimizing the estimated maintenance time. Each strategy includes operation and maintenance instructions for each power plant unit and the estimated base time for executing the corresponding instructions. The estimated completion time of each operation and maintenance instruction is calculated, and it is determined whether the time falls within the abnormal period of the corresponding power plant unit. If it falls within the abnormal period, the estimated base time of the corresponding instruction is adjusted based on the historical average health stress index of the power plant unit during that period. If it does not fall within the abnormal period, no adjustment is required.

[0009] The optimization strategy iterates through all maintenance actions, calculates the estimated execution time of each maintenance action after correction, and counts the total number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, the matching value between the strategy and the current overall state of the power plant is calculated. After sorting the matching values ​​from high to low, the strategy with the highest matching value is selected as the final recommended strategy.

[0010] To obtain historical operating data of the target photovoltaic power plant's units to be optimized, and to construct datasets for each power plant unit, the specific steps include:

[0011] Obtain historical operating data for all power plant units to be optimized in the target photovoltaic power plant, and generate a historical operating dataset {d1, d2, ..., d...} for the power plant units. i , ..., d I}, where d i This represents the historical operating data set of the i-th power station unit in the target photovoltaic power station, and I represents the number of power station units to be analyzed and optimized in the target photovoltaic power station. The power station unit includes strings and inverters.

[0012] Let d be the set of historical operating data of any selected power plant unit. a The historical operating data set includes historical power generation p aj Compared to historical performance, pr aj , where p aj pr represents the historical average output power of power plant unit a in the j-th statistical period; aj The historical performance ratio of power plant unit a in the j-th statistical period is represented by the ratio of the actual power generation to the theoretical maximum power generation in that period, where j = 1, 2, ..., J, and J represents the total number of statistical periods within the target analysis period.

[0013] Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit. Specific steps include:

[0014] The health stress index of power plant unit a in the j-th time period is calculated based on historical power generation and historical performance ratio, and is defined as follows: H aj =[(p aj / pr aj -p aj ] / C a Among them, H aj C represents the health stress index of power plant unit a in the j-th time period. a This indicates the rated capacity of power plant unit a;

[0015] The periods in which the health stress index is higher than the health threshold for N consecutive statistical periods are marked to form the historical health abnormality period set T of power plant unit i. i ={t i1 , t i2 , ..., t ij’ , ..., t iJ’}, where t ij’ This indicates that the i-th power plant unit was identified as abnormal during the j'-th time period, where J' represents the total number of time periods during which the unit was detected as having an abnormal health stress index. The health threshold is defined as the p-quantile of all health stress indices of power plant unit a within the historical statistical period, where p represents a preset percentage value.

[0016] Based on real-time operation data of the photovoltaic power plant and the goal of minimizing the estimated maintenance time, candidate optimization strategies are generated. These strategies include the operation and maintenance instructions for each power plant unit and the estimated base time for executing those instructions. The estimated completion time for each operation and maintenance instruction is calculated, and it is determined whether this time falls within a corresponding abnormal period. If it falls within an abnormal period, the estimated base time is adjusted based on the historical average health stress index of the power plant unit during that abnormal period; otherwise, no adjustment is made. Specific steps include:

[0017] Based on the real-time operation data of the target photovoltaic power plant and with the shortest expected maintenance time as the preset optimization objective, C candidate optimization strategies are generated. An arbitrary optimization strategy is selected and denoted as Q. c The optimization strategy includes a sequence of operation and maintenance instructions A for all power station units of the target photovoltaic power station. ci and the estimated base duration Y required to execute each instruction ci Where C represents the number of generated candidate optimization strategies, and A ci Y represents the operation and maintenance instruction for the i-th power station unit in the target photovoltaic power station under the c-th strategy scheme. ci This represents the estimated base time required to execute the i-th power plant unit in the c-th strategy scheme;

[0018] For strategy Q c Each operation and maintenance instruction A in ci Its expected completion time is te ci =t0+Σ k=1i Y ck , when te ci ∈T i When, t0 represents strategy Q c The starting time point of execution, te ci This indicates that for policy Q c Each operation and maintenance instruction A in ci The expected completion time;

[0019] The estimated actual execution time of this instruction is corrected using the following formula: Y1 ci =Y ci ×(1+α*avg1(H aj )); where α represents the proportion of loss in mapping the health stress index to execution efficiency, Y1 ci This represents the estimated time required to execute the i-th unit to be optimized in the modified target photovoltaic power plant under the c-th strategy scheme, avg1(H aj This represents the average historical health pressure index of power plant unit i during the abnormal period; otherwise, no correction is needed.

[0020] α is defined as follows: Based on the historical operation and maintenance work order dataset, multi-dimensional features including the historical health stress index of equipment, equipment type, operation and maintenance task type, and environmental time period characteristics are extracted. The ratio of actual execution time to planned time is used as the target variable. The relationship between the features and the target variable is trained using a machine learning regression algorithm to obtain a prediction model. The constructed prediction model is used to map the health stress index to the loss ratio of execution efficiency.

[0021] Iterate through all maintenance actions in the optimization strategy, calculate the estimated execution time of each action after correction, and count the number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, calculate the matching value between the strategy and the current overall state of the power plant. After sorting the matching values ​​from high to low, select the strategy with the highest matching value as the recommended strategy. The specific steps include:

[0022] Traversal optimization strategy Q c All maintenance actions in A ci Calculate the corrected estimated execution time and count all cases that satisfy Y1. ci >Y ci The number of instructions is calculated and accumulated to obtain K, where K represents the cumulative number of operations and maintenance actions whose revised estimated execution time is greater than the original estimated time among all operations and maintenance actions corresponding to the optimization strategy.

[0023] Calculate the optimization strategy Q c Matching value W with the current overall state of the power plant c The definition is as follows: Wc =1-[k / I], sort the matching values ​​from high to low, and select the optimization strategy with the highest matching value as the recommended strategy.

[0024] An IoT-based intelligent optimization system for photovoltaic power plants includes: a data acquisition module, an anomaly identification module, a strategy generation and duration correction module, and a strategy evaluation module. The data acquisition module acquires historical operational data of the target photovoltaic power plant's units to be optimized, constructing a dataset for each unit. The anomaly identification module calculates the health stress index for each statistical period, compares it with a preset threshold, and filters out periods exceeding the threshold, forming a set of historical health anomaly periods for the target power plant units. The strategy generation and duration correction module combines real-time operational data of the photovoltaic power plant to generate candidate optimization strategies with the shortest estimated maintenance time as the objective. Each strategy includes maintenance instructions for each power plant unit and an estimated base duration for executing the corresponding instructions. The estimated completion time of each maintenance instruction is calculated, and it is determined whether this time falls within the corresponding power plant unit's anomaly period. If it falls within an anomaly period, the estimated base duration of the corresponding instruction is corrected based on the average historical health stress index of the power plant unit during that period; otherwise, no correction is needed. The strategy evaluation module calculates the matching value between the strategy and the current state of the power plant, sorts the strategies by matching value, and recommends the optimal strategy.

[0025] The data acquisition module includes a historical operation data acquisition unit and a dataset construction unit. The historical operation data acquisition unit is used to collect historical data of the target photovoltaic power station unit to be optimized, including historical power generation and historical performance ratio for each statistical period. The dataset construction unit is used to classify and organize the collected historical data and generate corresponding historical operation datasets for each power station unit.

[0026] The anomaly identification module includes a health stress index calculation unit, a health threshold setting unit, and an abnormal period marking unit. The health stress index calculation unit is used to calculate the health stress index for each statistical period by combining the historical power generation, historical performance ratio, and rated capacity of the power plant unit. The health threshold setting unit is used to determine the health threshold based on all health stress indices within the historical statistical period of the power plant unit. The abnormal period marking unit is used to form a set of historical abnormal health periods for each power plant unit.

[0027] The strategy generation and duration correction module includes a candidate optimization strategy generation unit and an estimated duration correction unit. The candidate optimization strategy generation unit is used to generate candidate optimization strategies by combining real-time operation data of the photovoltaic power station with the goal of minimizing the estimated maintenance time. The estimated duration correction unit is used to determine whether the estimated completion time of the instruction is in an abnormal period, and to correct the original estimated duration by combining the historical average health stress index of the power station unit in that abnormal period. If it is not in an abnormal period, the original duration remains unchanged.

[0028] The strategy evaluation module includes a corrected time consumption statistics unit, a matching value calculation unit, and a recommendation unit. The corrected time consumption statistics unit is used to traverse all operation and maintenance actions in each candidate strategy, calculate the corrected estimated execution time, and count the total number of actions whose corrected time consumption exceeds the original estimated time. Based on the above statistical results, the matching value calculation unit calculates the matching value between the candidate strategy and the current overall state of the power plant. The recommendation unit sorts all candidate strategies from high to low according to the matching value and selects the strategy with the highest matching value as the final recommended strategy.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] 1. This invention introduces a dynamic correction mechanism for operation and maintenance duration, which adjusts the estimated duration based on the historical average health stress index during abnormal equipment periods. This differs from the static and fixed operation and maintenance duration settings in existing technologies, and can adapt to the actual health status of the equipment.

[0031] 2. This invention establishes an adaptability assessment system between strategies and the current state of the power plant. By statistically correcting the number of actions that take longer than expected, a matching value is calculated, and the optimal strategy is selected based on the matching value. Unlike existing technologies that lack targeted strategy assessment, this invention can select an operation and maintenance plan that fits the real-time state of the power plant. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating an intelligent optimization method for photovoltaic power plants based on the Internet of Things (IoT) according to the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] like Figure 1 As shown, the present invention provides a technical solution, a smart optimization method for photovoltaic power plants based on the Internet of Things, the method comprising the following steps:

[0035] Obtain historical operating data of the target photovoltaic power plant's units to be optimized, and construct datasets for each power plant unit;

[0036] Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out the periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit;

[0037] Based on real-time operation data of photovoltaic power plants, candidate optimization strategies are generated with the goal of minimizing the estimated maintenance time. Each strategy includes operation and maintenance instructions for each power plant unit and the estimated base time for executing the corresponding instructions. The estimated completion time of each operation and maintenance instruction is calculated, and it is determined whether the time falls within the abnormal period of the corresponding power plant unit. If it falls within the abnormal period, the estimated base time of the corresponding instruction is adjusted based on the historical average health stress index of the power plant unit during that period. If it does not fall within the abnormal period, no adjustment is required.

[0038] The optimization strategy iterates through all maintenance actions, calculates the estimated execution time of each maintenance action after correction, and counts the total number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, the matching value between the strategy and the current overall state of the power plant is calculated. After sorting the matching values ​​from high to low, the strategy with the highest matching value is selected as the final recommended strategy.

[0039] To obtain historical operating data of the target photovoltaic power plant's units to be optimized, and to construct datasets for each power plant unit, the specific steps include:

[0040] Obtain historical operating data for all power plant units to be optimized in the target photovoltaic power plant, and generate a historical operating dataset {d1, d2, ..., d...} for the power plant units. i , ..., d I}, where d i This represents the historical operating data set of the i-th power station unit in the target photovoltaic power station, and I represents the number of power station units to be analyzed and optimized in the target photovoltaic power station. The power station unit includes strings and inverters.

[0041] Let d be the set of historical operating data of any selected power plant unit. a The historical operating data set includes historical power generation p aj Compared to historical performance, pr aj , where p aj pr represents the historical average output power of power plant unit a in the j-th statistical period; aj The historical performance ratio of power plant unit a in the j-th statistical period is represented by the ratio of the actual power generation to the theoretical maximum power generation in that period, where j = 1, 2, ..., J, and J represents the total number of statistical periods within the target analysis period.

[0042] Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit. Specific steps include:

[0043] The health stress index of power plant unit a in the j-th time period is calculated based on historical power generation and historical performance ratio, and is defined as follows: H aj =[(p aj / pr aj -p aj ] / C a Among them, H aj C represents the health stress index of power plant unit a in the j-th time period. a This indicates the rated capacity of power plant unit a;

[0044] The periods in which the health stress index is higher than the health threshold for N consecutive statistical periods are marked to form the historical health abnormality period set T of power plant unit i. i ={t i1 , t i2 , ..., t ij’ , ..., t iJ’}, where t ij’ This indicates that the i-th power plant unit was identified as abnormal during the j'-th time period, where J' represents the total number of time periods during which the unit was detected as having an abnormal health stress index. The health threshold is defined as the p-quantile of all health stress indices of power plant unit a within the historical statistical period, where p represents a preset percentage value.

[0045] Based on real-time operation data of the photovoltaic power plant and the goal of minimizing the estimated maintenance time, candidate optimization strategies are generated. These strategies include the operation and maintenance instructions for each power plant unit and the estimated base time for executing those instructions. The estimated completion time for each operation and maintenance instruction is calculated, and it is determined whether this time falls within a corresponding abnormal period. If it falls within an abnormal period, the estimated base time is adjusted based on the historical average health stress index of the power plant unit during that abnormal period; otherwise, no adjustment is made. Specific steps include:

[0046] Based on the real-time operation data of the target photovoltaic power plant and with the shortest expected maintenance time as the preset optimization objective, C candidate optimization strategies are generated. An arbitrary optimization strategy is selected and denoted as Q. c The optimization strategy includes a sequence of operation and maintenance instructions A for all power station units of the target photovoltaic power station. ci and the estimated base duration Y required to execute each instruction ci Where C represents the number of generated candidate optimization strategies, and A ciY represents the operation and maintenance instruction for the i-th power station unit in the target photovoltaic power station under the c-th strategy scheme. ci This represents the estimated base time required to execute the i-th power plant unit in the c-th strategy scheme;

[0047] For strategy Q c Each operation and maintenance instruction A in ci Its expected completion time is te ci =t0+Σ k=1 i Y ck , when te ci ∈T i When, t0 represents strategy Q c The starting time point of execution, te ci This indicates that for policy Q c Each operation and maintenance instruction A in ci The expected completion time;

[0048] The estimated actual execution time of this instruction is corrected using the following formula: Y1 ci =Y ci ×(1+α*avg1(H aj )); where α represents the proportion of loss in mapping the health stress index to execution efficiency, Y1 ci This represents the estimated time required to execute the i-th unit to be optimized in the modified target photovoltaic power plant under the c-th strategy scheme, avg1(H aj This represents the average historical health pressure index of power plant unit i during the abnormal period; otherwise, no correction is needed.

[0049] α is defined as follows: Based on the historical operation and maintenance work order dataset, multi-dimensional features including the historical health stress index of equipment, equipment type, operation and maintenance task type, and environmental time period characteristics are extracted. The ratio of actual execution time to planned time is used as the target variable. The relationship between the features and the target variable is trained using a machine learning regression algorithm to obtain a prediction model. The constructed prediction model is used to map the health stress index to the loss ratio of execution efficiency.

[0050] Iterate through all maintenance actions in the optimization strategy, calculate the estimated execution time of each action after correction, and count the number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, calculate the matching value between the strategy and the current overall state of the power plant. After sorting the matching values ​​from high to low, select the strategy with the highest matching value as the recommended strategy. The specific steps include:

[0051] Traversal optimization strategy Q c All maintenance actions in A ciCalculate the corrected estimated execution time and count all cases that satisfy Y1. ci >Y ci The number of instructions is calculated and accumulated to obtain K, where K represents the cumulative number of operations and maintenance actions whose revised estimated execution time is greater than the original estimated time among all operations and maintenance actions corresponding to the optimization strategy.

[0052] Calculate the optimization strategy Q c Matching value W with the current overall state of the power plant c The definition is as follows: W c =1-[k / I], sort the matching values ​​from high to low, and select the optimization strategy with the highest matching value as the recommended strategy.

[0053] In Example 1: Historical operating data of all units to be optimized in the power plant are collected through IoT sensing devices, covering key information such as power generation and performance of each unit in different operating periods. The performance is determined by the ratio of actual power generation to theoretical optimal power generation. The staff classifies and organizes the data, and builds a dedicated historical operating dataset for each power plant unit to ensure that the operating trajectory of each unit can be completely recorded, providing a comprehensive basis for subsequent analysis.

[0054] By combining the historical power generation, performance and rated capacity of each unit, the health stress index of each operating period is calculated to reflect the operating load and health status of the unit. Based on all health stress indices in the historical operating cycle of each unit, a reasonable health threshold is set. This threshold can distinguish the critical standard between normal operation and abnormal state. Through continuous monitoring, the periods when the health stress index exceeds the threshold are marked to form a record of the historical health abnormal period of each unit, and to identify the specific periods when each unit is prone to operating abnormalities.

[0055] Real-time data collection of the power plant's current operation, including overall power generation status, equipment operating status, ambient light conditions, component temperature, energy storage equipment status, and scheduling requirements, is used to generate several candidate operation and maintenance (O&M) plans with the shortest maintenance time as the core objective. Each plan includes specific O&M operation instructions for all power plant units, as well as the estimated base time required to execute each instruction. Based on the start time of the plan and the estimated duration of previous O&M operations, the estimated completion time of each instruction is calculated. This time is then compared with the historical health anomaly period records of the corresponding unit to determine if it falls within an anomaly period. If the estimated completion time falls within an anomaly period, the average historical health stress index of the unit during that period is retrieved, and the original estimated base time is reasonably adjusted based on the mapping relationship obtained through machine learning training. If it does not fall within an anomaly period, the original estimated time remains unchanged.

[0056] The process involves iterating through all the operation and maintenance operations in each candidate solution, calculating the estimated execution time of each operation after correction, and counting the number of operations whose time exceeds the original estimated time after correction. Based on this statistical result, the degree of matching between the candidate solution and the current overall operating status of the power plant is calculated. Then, all candidate solutions are sorted from high to low according to the degree of matching, and finally, the solution with the highest degree of matching is selected as the optimal operation and maintenance strategy and recommended to the staff.

[0057] An IoT-based intelligent optimization system for photovoltaic power plants includes: a data acquisition module, an anomaly identification module, a strategy generation and duration correction module, and a strategy evaluation module. The data acquisition module acquires historical operational data of the target photovoltaic power plant's units to be optimized, constructing a dataset for each unit. The anomaly identification module calculates the health stress index for each statistical period, compares it with a preset threshold, and filters out periods exceeding the threshold, forming a set of historical health anomaly periods for the target power plant units. The strategy generation and duration correction module combines real-time operational data of the photovoltaic power plant to generate candidate optimization strategies with the shortest estimated maintenance time as the objective. Each strategy includes maintenance instructions for each power plant unit and an estimated base duration for executing the corresponding instructions. The estimated completion time of each maintenance instruction is calculated, and it is determined whether this time falls within the corresponding power plant unit's anomaly period. If it falls within an anomaly period, the estimated base duration of the corresponding instruction is corrected based on the average historical health stress index of the power plant unit during that period; otherwise, no correction is needed. The strategy evaluation module calculates the matching value between the strategy and the current state of the power plant, sorts the strategies by matching value, and recommends the optimal strategy.

[0058] The data acquisition module includes a historical operation data acquisition unit and a dataset construction unit. The historical operation data acquisition unit is used to collect historical data of the target photovoltaic power station unit to be optimized, including historical power generation and historical performance ratio for each statistical period. The dataset construction unit is used to classify and organize the collected historical data and generate corresponding historical operation datasets for each power station unit.

[0059] The anomaly identification module includes a health stress index calculation unit, a health threshold setting unit, and an abnormal period marking unit. The health stress index calculation unit is used to calculate the health stress index for each statistical period by combining the historical power generation, historical performance ratio, and rated capacity of the power plant unit. The health threshold setting unit is used to determine the health threshold based on all health stress indices within the historical statistical period of the power plant unit. The abnormal period marking unit is used to form a set of historical abnormal health periods for each power plant unit.

[0060] The strategy generation and duration correction module includes a candidate optimization strategy generation unit and an estimated duration correction unit. The candidate optimization strategy generation unit is used to generate candidate optimization strategies by combining real-time operation data of the photovoltaic power station with the goal of minimizing the estimated maintenance time. The estimated duration correction unit is used to determine whether the estimated completion time of the instruction is in an abnormal period, and to correct the original estimated duration by combining the historical average health stress index of the power station unit in that abnormal period. If it is not in an abnormal period, the original duration remains unchanged.

[0061] The strategy evaluation module includes a corrected time consumption statistics unit, a matching value calculation unit, and a recommendation unit. The corrected time consumption statistics unit is used to traverse all operation and maintenance actions in each candidate strategy, calculate the corrected estimated execution time, and count the total number of actions whose corrected time consumption exceeds the original estimated time. Based on the above statistical results, the matching value calculation unit calculates the matching value between the candidate strategy and the current overall state of the power plant. The recommendation unit sorts all candidate strategies from high to low according to the matching value and selects the strategy with the highest matching value as the final recommended strategy.

[0062] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A smart optimization method for photovoltaic power plants based on the Internet of Things, characterized in that: The method includes the following steps: Obtain historical operating data of the target photovoltaic power plant's units to be optimized, and construct datasets for each power plant unit; Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out the periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit; Based on real-time operation data of photovoltaic power plants, candidate optimization strategies are generated with the goal of minimizing the estimated maintenance time. Each strategy includes operation and maintenance instructions for each power plant unit and the estimated base time for executing the corresponding instructions. The estimated completion time of each operation and maintenance instruction is calculated, and it is determined whether the time falls within the abnormal period of the corresponding power plant unit. If it falls within the abnormal period, the estimated base time of the corresponding instruction is adjusted based on the historical average health stress index of the power plant unit during that period. If it does not fall within the abnormal period, no adjustment is required. The optimization strategy iterates through all maintenance actions, calculates the estimated execution time of each maintenance action after correction, and counts the total number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, the matching value between the strategy and the current overall state of the power plant is calculated. After sorting the matching values ​​from high to low, the strategy with the highest matching value is selected as the final recommended strategy.

2. The intelligent optimization method for photovoltaic power plants based on the Internet of Things according to claim 1, characterized in that: To obtain historical operating data of the target photovoltaic power plant's units to be optimized, and to construct datasets for each power plant unit, the specific steps include: Obtain historical operating data for all power plant units to be optimized in the target photovoltaic power plant, and generate a historical operating dataset {d1, d2, ..., d...} for the power plant units. i , ..., d I }, where d i This represents the historical operating data set of the i-th power station unit in the target photovoltaic power station, and I represents the number of power station units to be analyzed and optimized in the target photovoltaic power station. The power station unit includes strings and inverters. Let d be the set of historical operating data of any selected power plant unit. a The historical operating data set includes historical power generation p aj Compared to historical performance, pr aj , where p aj pr represents the historical average output power of power plant unit a in the j-th statistical period; aj The historical performance ratio of power plant unit a in the j-th statistical period is represented by the ratio of the actual power generation to the theoretical maximum power generation in that period, where j = 1, 2, ..., J, and J represents the total number of statistical periods within the target analysis period.

3. The intelligent optimization method for photovoltaic power plants based on the Internet of Things according to claim 2, characterized in that: Calculate the health stress index for each statistical period, compare it with a preset threshold, and filter out periods that exceed the threshold to form a set of historical health abnormality periods for the target power plant unit. Specific steps include: The health stress index of power plant unit a in the j-th time period is calculated based on historical power generation and historical performance ratio, and is defined as follows: H aj =[(p aj / pr aj -p aj ] / C a ; where H aj C represents the health stress index of power plant unit a in the j-th time period. a This indicates the rated capacity of power plant unit a; The periods in which the health stress index is higher than the health threshold for N consecutive statistical periods are marked to form the historical health abnormality period set T of power plant unit i. i ={t i1 , t i2 , ..., t ij’ , ..., t iJ’ }, where t ij’ This indicates that the i-th power plant unit was identified as abnormal during the j'-th time period, where J' represents the total number of time periods during which the unit was detected as having an abnormal health stress index. The health threshold is defined as the p-quantile of all health stress indices of power plant unit a within the historical statistical period, where p represents a preset percentage value.

4. The intelligent optimization method for photovoltaic power plants based on the Internet of Things according to claim 3, characterized in that: Based on real-time operation data of the photovoltaic power plant and the goal of minimizing the estimated maintenance time, candidate optimization strategies are generated. These strategies include the operation and maintenance instructions for each power plant unit and the estimated base time for executing those instructions. The estimated completion time for each operation and maintenance instruction is calculated, and it is determined whether this time falls within a corresponding abnormal period. If it falls within an abnormal period, the estimated base time is adjusted based on the historical average health stress index of the power plant unit during that abnormal period; otherwise, no adjustment is made. Specific steps include: Based on the real-time operation data of the target photovoltaic power plant and with the shortest expected maintenance time as the preset optimization objective, C candidate optimization strategies are generated. An arbitrary optimization strategy is selected and denoted as Q. c The optimization strategy includes a sequence of operation and maintenance instructions A for all power station units of the target photovoltaic power station. ci and the estimated base duration Y required to execute each instruction ci Where C represents the number of generated candidate optimization strategies, and A ci Y represents the operation and maintenance instruction for the i-th power station unit in the target photovoltaic power station under the c-th strategy scheme. ci This represents the estimated base time required to execute the i-th power plant unit in the c-th strategy scheme; For strategy Q c Each operation and maintenance instruction A in ci Its expected completion time is te ci =t0+Σ k=1 i Y ck , when te ci ∈T i When, t0 represents strategy Q c The starting time point of execution, te ci This indicates that for policy Q c Each operation and maintenance instruction A in ci The expected completion time; The estimated actual execution time of this instruction is corrected using the following formula: Y1 ci =Y ci ×(1+α*avg1(H aj )); where α represents the proportion of loss in mapping the health stress index to execution efficiency, Y1 ci This represents the estimated time required to execute the i-th unit to be optimized in the modified target photovoltaic power plant under the c-th strategy scheme, avg1(H aj This represents the average historical health pressure index of power plant unit i during the abnormal period; otherwise, no correction is needed. α is defined as follows: Based on the historical operation and maintenance work order dataset, multi-dimensional features including the historical health stress index of equipment, equipment type, operation and maintenance task type, and environmental time period characteristics are extracted. The ratio of actual execution time to planned time is used as the target variable. The relationship between the features and the target variable is trained using a machine learning regression algorithm to obtain a prediction model. The constructed prediction model is used to map the health stress index to the loss ratio of execution efficiency.

5. The intelligent optimization method for photovoltaic power plants based on the Internet of Things according to claim 4, characterized in that: Iterate through all maintenance actions in the optimization strategy, calculate the estimated execution time of each action after correction, and count the number of actions whose corrected execution time exceeds the original estimated time. Based on the statistical results, calculate the matching value between the strategy and the current overall state of the power plant. After sorting the matching values ​​from high to low, select the strategy with the highest matching value as the recommended strategy. The specific steps include: Traversal optimization strategy Q c All maintenance actions in A ci Calculate the corrected estimated execution time and count all cases that satisfy Y1. ci >Y ci The number of instructions is calculated and accumulated to obtain K, where K represents the cumulative number of operations and maintenance actions whose revised estimated execution time is greater than the original estimated time among all operations and maintenance actions corresponding to the optimization strategy. Calculate the optimization strategy Q c Matching value W with the current overall state of the power plant c The definition is as follows: W c =1-[k / I], sort the matching values ​​from high to low, and select the optimization strategy with the highest matching value as the recommended strategy.

6. An IoT-based intelligent optimization system for photovoltaic power plants, applied to the IoT-based intelligent optimization method for photovoltaic power plants as described in any one of claims 1-5, characterized in that: The system includes: a data acquisition module, an anomaly identification module, a strategy generation and duration correction module, and a strategy evaluation module. The data acquisition module acquires historical operational data of the target photovoltaic power plant's units to be optimized, constructing a dataset for each unit. The anomaly identification module calculates the health stress index for each statistical period, compares it with a preset threshold, and filters out periods exceeding the threshold, forming a set of historical health anomaly periods for the target power plant units. The strategy generation and duration correction module combines real-time operational data of the photovoltaic power plant to generate candidate optimization strategies with the shortest estimated maintenance time as the objective. Each strategy includes maintenance instructions for each power plant unit and an estimated base duration for executing the corresponding instructions. It calculates the estimated completion time of each maintenance instruction and determines whether this time falls within the corresponding power plant unit's anomaly period. If it falls within an anomaly period, the estimated base duration of the corresponding instruction is corrected based on the average historical health stress index of the power plant unit during that period; otherwise, no correction is needed. The strategy evaluation module calculates the matching value between the strategy and the current state of the power plant, sorts the strategies by matching value, and recommends the optimal strategy.

7. The intelligent optimization system for a photovoltaic power station based on the Internet of Things according to claim 6, characterized in that: The data acquisition module includes a historical operation data acquisition unit and a dataset construction unit. The historical operation data acquisition unit is used to collect historical data of the target photovoltaic power station unit to be optimized, including historical power generation and historical performance ratio for each statistical period. The dataset construction unit is used to classify and organize the collected historical data and generate corresponding historical operation datasets for each power station unit.

8. The intelligent optimization system for a photovoltaic power station based on the Internet of Things according to claim 7, characterized in that: The anomaly identification module includes a health stress index calculation unit, a health threshold setting unit, and an abnormal period marking unit. The health stress index calculation unit is used to calculate the health stress index for each statistical period by combining the historical power generation, historical performance ratio, and rated capacity of the power plant unit. The health threshold setting unit is used to determine the health threshold based on all health stress indices within the historical statistical period of the power plant unit. The abnormal time period marking unit is used to form a set of historical health abnormal time periods for each power plant unit.

9. The intelligent optimization system for a photovoltaic power station based on the Internet of Things according to claim 8, characterized in that: The strategy generation and duration correction module includes a candidate optimization strategy generation unit and an estimated duration correction unit. The candidate optimization strategy generation unit is used to generate candidate optimization strategies by combining real-time operation data of the photovoltaic power plant with the goal of minimizing the estimated maintenance time. The estimated duration correction unit is used to determine if the estimated completion time of the instruction is in an abnormal period, and then, in conjunction with the historical average health pressure index of the power plant unit in that abnormal period, correct the original estimated base duration. If it is not during an abnormal period, the original duration will remain unchanged.

10. The intelligent optimization system for a photovoltaic power station based on the Internet of Things according to claim 9, characterized in that: The strategy evaluation module includes a corrected time consumption statistics unit, a matching value calculation unit, and a recommendation unit. The corrected time consumption statistics unit is used to traverse all operation and maintenance actions in each candidate strategy, calculate the corrected estimated execution time, and count the total number of actions whose corrected time consumption exceeds the original estimated time. Based on the above statistical results, the matching value calculation unit calculates the matching value between the candidate strategy and the current overall state of the power plant; the recommendation unit sorts all candidate strategies from high to low according to the matching value and selects the strategy with the highest matching value as the final recommended strategy.

Images