A photovoltaic equipment full life cycle operation and maintenance management system

Abstract

This invention discloses a photovoltaic (PV) equipment lifecycle operation and maintenance (O&M) management system, relating to the field of PV equipment O&M management technology. It includes: a physical state perception and event input module for collecting physical parameters of external events associated with the PV equipment and obtaining real-time status data of the PV equipment; a physical failure prediction model module for simulating the damage accumulation process of the PV equipment and calculating a failure probability map characterizing the future failure likelihood of each device; an economic scheduling and O&M decision-making module for solving the optimal predictive maintenance strategy by comprehensively considering preset economic constraints and aiming to minimize the generalized risk cost of the PV equipment's entire lifecycle; and a system execution and feedback module for generating O&M work orders and providing feedback to calibrate the physical failure prediction model module. This invention transforms O&M management into proactive management based on risk prediction, enabling the simulation of future risk evolution states of the power plant and proactive intervention.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic equipment operation and maintenance management technology, and in particular to a photovoltaic equipment full life cycle operation and maintenance management system. Background Technology

[0002] In the traditional operation and maintenance management of photovoltaic power plants, a passive response maintenance mode is usually adopted, that is, intervention is only carried out after obvious equipment failure. This mode cannot effectively predict and manage the long-term latent gradual equipment damage triggered by external physical events such as hail. In addition, existing technical solutions often separate the model for predicting the physical state of equipment from the economic model for operation and maintenance decision-making, resulting in a lack of accurate consideration of the actual risk state of the equipment in the future operation and maintenance decision-making, making it difficult to maximize asset value from the perspective of the entire life cycle.

[0003] Therefore, the industry urgently needs an advanced technology that can integrate physical damage prediction with economic benefit analysis and realize the shift from passive maintenance to proactive risk management. Summary of the Invention

[0004] The purpose of this invention is to provide a photovoltaic equipment full lifecycle operation and maintenance management system to solve the above-mentioned technical problems, which is achieved through the following technical solution:

[0005] A photovoltaic equipment full lifecycle operation and maintenance management system includes:

[0006] The physical state sensing and event input module is used to collect physical parameters of external events associated with the photovoltaic equipment and obtain real-time status data of the photovoltaic equipment.

[0007] The physical failure prediction model module is used to deduce the damage accumulation process of the photovoltaic equipment based on the physical parameters and real-time status data collected by the physical state perception and event input module, and calculate the failure probability map characterizing the future failure possibility of each device.

[0008] The economic scheduling and operation and maintenance decision module is used to receive the failure probability map calculated by the physical failure prediction model module, and with the goal of minimizing the generalized risk cost of the photovoltaic equipment throughout its entire life cycle, and taking into account the preset economic constraints, solve for the optimal predictive maintenance strategy.

[0009] The system execution and feedback module is used to generate maintenance work orders based on the optimal predictive maintenance strategy solved by the economic scheduling and maintenance decision module, collect the actual results after maintenance execution, and use the actual results to calibrate the physical failure prediction model module.

[0010] Preferably, the physical failure prediction model module is further configured to: determine the instantaneous damage increment caused by the external event, and determine the progressive damage increment caused by daily environmental stress;

[0011] Combining the instantaneous damage increment and the progressive damage increment, and based on the damage degree at the previous moment, the current potential damage degree, which characterizes the degree of microscopic damage inside the device, is iteratively updated.

[0012] Preferably, the instantaneous damage increment is determined based on the impact energy and impact incident angle of the external event, combined with the impact area of ​​the equipment and a preset material fragility coefficient characterizing the equipment's ability to resist physical impact.

[0013] Preferably, the determination of the incremental damage is based on the collected daily temperature cycle range and the daily average ultraviolet index, and the application of a damage acceleration factor; wherein, the damage acceleration factor is used to characterize the accelerating effect on the subsequent daily aging process determined by the current potential damage level.

[0014] Preferably, the physical failure prediction model module is further configured to: take the current potential damage level as input, apply a preset S-shaped function for transformation, and generate the failure probability map; wherein, the transformation process of the S-shaped function applies a preset critical damage threshold.

[0015] Preferably, the economic scheduling and operation and maintenance decision module is further configured to: determine the direct costs incurred by executing the operation and maintenance decision, and determine the additional expected risk costs caused by not executing the operation and maintenance decision; combine the direct costs and the additional expected risk costs, and after time value discounting, construct the objective function of the operation and maintenance scheduling model that minimizes the risk costs.

[0016] Preferably, the determination of the additional expected risk cost is based on the calculation of the additional failure probability and the total economic loss caused by a single failure within the current time interval; wherein, the additional failure probability is derived from the time change of the future failure probability in the failure probability map.

[0017] Preferably, the system execution and feedback module is further configured to: update the current potential damage level in response to the execution of the maintenance work order;

[0018] When the maintenance work order type is "replacement", the current potential damage level is reset to the preset factory initial value.

[0019] When the type of the maintenance work order is repair, the current potential damage level is reset to a value determined based on the repair effect evaluation.

[0020] Compared with existing technologies, the photovoltaic equipment full life cycle operation and maintenance management system provided by the present invention has the following beneficial effects:

[0021] 1. The system transforms operation and maintenance management from reactive fault response to proactive management based on risk prediction. It can simulate the future risk evolution of the power plant and make forward-looking interventions, rather than just dealing with faults that have already occurred, thereby effectively avoiding potential major losses.

[0022] 2. This system overcomes the shortcomings of traditional technologies where physical models and economic models are disconnected. By quantifying the progressive physical damage process into probabilistic risks that can be used for economic decision-making, it establishes a key data link between the evolution of equipment physical state and economic cost-benefit analysis, making operation and maintenance decisions more scientific and accurate.

[0023] 3. This system can not only distinguish between instantaneous damage caused by external impacts and gradual aging caused by daily environmental stresses, but also introduces damage acceleration effects and simulates the nonlinear characteristics of damage accumulation. This enables it to proactively identify and manage long-latency equipment damage caused by external events, significantly improving the physical realism and accuracy of the predictions.

[0024] 4. The system can translate optimal operation and maintenance decisions into work order execution and collect the actual results after execution, such as the actual damage status after component replacement. This real data will be fed back to calibrate and optimize the prediction model, forming a dynamic feedback loop that ensures the system can learn and evolve on its own, continuously improving the accuracy and effectiveness of full lifecycle management. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a module of a photovoltaic equipment full life cycle operation and maintenance management system according to the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0027] Example 1

[0028] Please see Figure 1 A photovoltaic equipment full lifecycle operation and maintenance management system, comprising:

[0029] The physical state sensing and event input module is used to collect physical parameters of external events associated with photovoltaic equipment and obtain real-time status data of photovoltaic equipment.

[0030] The physical failure prediction model module is used to deduce the damage accumulation process of photovoltaic equipment based on the physical parameters and real-time status data collected by the physical state perception and event input module, and calculate the failure probability map that represents the future failure possibility of each device.

[0031] The economic scheduling and operation and maintenance decision module is used to receive the failure probability map calculated by the physical failure prediction model module. With the goal of minimizing the generalized risk cost of the entire life cycle of photovoltaic equipment, and taking into account the preset economic constraints, it solves the optimal predictive maintenance strategy.

[0032] The system execution and feedback module is used to generate maintenance work orders based on the optimal predictive maintenance strategy solved by the economic scheduling and operation and maintenance decision module, and to collect the actual results after the operation and maintenance is executed. The actual results are then used to calibrate the physical failure prediction model module.

[0033] This embodiment provides a photovoltaic equipment full life cycle operation and maintenance management system; the system aims to solve the technical problem that existing technologies, due to model fragmentation and methodological defects, cannot effectively predict and manage the progressive and systemic equipment risks triggered by external physical events through a closed-loop process from physical damage prediction to economic decision optimization;

[0034] The system's workflow begins with the physical state perception and event input module. This module aims to provide the necessary and accurate initial data input for subsequent physical failure prediction. In this embodiment, it is implemented by integrating a meteorological data interface with the power plant monitoring and data acquisition system. For example, after a hailstorm, this module actively collects and structures the physical parameters of the event, such as the average diameter, density, terminal velocity, and impact area of ​​the hailstones. Simultaneously, it continuously acquires real-time status data of the photovoltaic equipment, such as temperature data collected by temperature sensors mounted on the module surface and irradiance data collected by ultraviolet sensors.

[0035] After acquiring the data, the system activates the physical failure prediction model module. The core purpose of this module is to quantitatively deduce the accumulation process of microscopic damage inside photovoltaic equipment based on the input physical parameters and state data, and to transform this physical process into a probabilistic future risk indicator that has direct guiding significance for decision-making. In this embodiment, the module calculates the failure probability of each affected photovoltaic device at various future time points by running a potential damage accumulation model based on the principles of damage mechanics, thereby generating a dynamic and forward-looking failure probability map. This map is the key data link for establishing the logical connection between the evolution of the physical world and economic decision analysis.

[0036] The economic scheduling and operation and maintenance decision-making module receives the aforementioned fault probability map as its core input. The function of this module is to transform operation and maintenance decisions from a traditional cost control problem into an investment decision problem oriented towards maximizing risk aversion value. In this embodiment, the module embeds a risk-cost minimization operation and maintenance scheduling model based on operations research theory. This model takes minimizing the generalized risk cost of the photovoltaic equipment throughout its entire planning cycle as its sole optimization objective, while also incorporating real-world economic constraints, such as the annual operation and maintenance budget ceiling, the quantity of available spare parts inventory, and the maximum daily working hours of the maintenance team, as constraints. By solving this optimization model, the module can inversely calculate the optimal predictive maintenance strategy that, under current resource constraints, can minimize potential future losses. This strategy is manifested as operation and maintenance instructions for specific equipment at specific times, for example, performing a replacement operation on module 15 in area A on day 30.

[0037] The system execution and feedback module is responsible for executing the optimal strategy and forming a feedback loop. Its purpose is to ensure that the optimal strategy is effectively executed and to use the execution results to optimize the model in reverse, thereby improving the long-term prediction accuracy of the system. In this embodiment, the module first automatically converts the optimal predictive maintenance strategy into a standardized operation and maintenance work order and sends it to the mobile terminal of the on-site operation and maintenance team. After the operation and maintenance activity is executed, the module will collect the actual results after the operation and maintenance is executed. For example, by performing electroluminescence detection on the replaced component, the true damage status of its internal microcracks can be obtained. These real physical damage data will be fed back to the physical failure prediction model module for calibration and optimization of key parameters in the model, such as the energy-damage conversion coefficient or the material fragility coefficient.

[0038] This embodiment constructs an adaptive optimization closed loop of perception-prediction-decision-execution-feedback through the coordinated work of the four modules mentioned above. It overcomes the defect of the physical model and economic model in the prior art, quantifies the gradual physical damage process into a probabilistic risk that can be used for economic decision-making, and realizes the transformation from reactive post-event maintenance to proactive management based on risk prediction. The system can proactively identify and manage long-latency equipment damage caused by external events, and transform its operation and maintenance decision-making function from simply dealing with the faults that have occurred to simulating the future risk state of the power plant and making proactive interventions, thereby maximizing the asset value of the power plant throughout its entire life cycle while ensuring the long-term stable operation of the power plant.

[0039] Example 2

[0040] The physical failure prediction model module is also used to: determine the instantaneous damage increment caused by external events and the progressive damage increment caused by daily environmental stresses;

[0041] By combining instantaneous damage increment and progressive damage increment, and based on the damage degree at the previous moment, the current potential damage degree, which characterizes the degree of microscopic damage inside the device, is iteratively updated and generated.

[0042] The instantaneous damage increment is determined based on the impact energy and impact incident angle of the external event, combined with the impact area of ​​the equipment and the preset material fragility coefficient that characterizes the equipment's ability to resist physical impact.

[0043] The incremental damage is determined based on the temperature cycle range and the average UV index of the day, and by applying a damage acceleration factor. The damage acceleration factor is used to characterize the accelerating effect on the subsequent daily aging process determined by the current potential damage level.

[0044] The physical failure prediction model module is also used to: take the current potential damage level as input, apply a preset S-shaped function for transformation, and generate a failure probability map; wherein, the transformation process of the S-shaped function applies a preset critical damage threshold.

[0045] This embodiment is a further explanation of Embodiment 1, specifying the specific implementation of the physical failure prediction model module. The core of this module is to iteratively update a current potential damage degree that characterizes the degree of microscopic damage inside the equipment. The update process combines the instantaneous damage increment caused by external events and the progressive damage increment caused by daily environmental stress. The instantaneous damage increment is determined based on impact energy, angle, and equipment parameters; the progressive damage increment is determined based on environmental factors and by applying a damage acceleration factor. The potential damage degree is converted into a failure probability through a S-shaped function.

[0046] The core of the physical failure prediction model module is a key internal state variable—potential damage level. Perform precise iterative calculations;

[0047] Potential damage This refers to a method used to quantitatively characterize the first... A photovoltaic device in A dimensionless parameter representing the degree of accumulation of microscopic damage that exists within a given timeframe but is macroscopically invisible; its function is to abstract the complex physical damage process into a calculable and iterable numerical value; it is derived from the damage accumulation model described below, based on the damage degree of the previous timeframe and various damage increments at the current timeframe. This module updates the potential damage degree through the following core evolution process:

[0048] ;

[0049] This formula shows that the total damage at the current moment is the damage at the previous moment plus two different types of new damage: instantaneous damage increment and progressive damage increment.

[0050] Instantaneous damage increment Its purpose is to quantify the instantaneous, one-time physical damage to equipment caused by external sudden events (such as hail or rock impacts); in this embodiment, the determination method is based on the principles of energy conservation and impact dynamics, and the calculation formula is as follows:

[0051]

[0052] in: It represents the average kinetic energy of impacting objects such as hail, with the dimension of joule (J). Its value is calculated by the physical state perception and event input module based on meteorological data. It represents the angle of impact, with the dimension in radians, and its value is determined based on the installation tilt angle of the photovoltaic module and the trajectory of the impacting object; It represents the impact area of ​​the component, measured in square meters, and its value comes from the equipment's factory parameters; The material fragility coefficient is a characterization factor of the first material. The dimensionless key parameter of a device’s ability to resist physical impact reflects the differences in impact resistance performance of components of different technologies and manufacturers. Its initial value is calibrated by static or dynamic loading destructive test data of different types of components, and can be optimized online by real damage data fed back by the system execution and feedback module. The energy-damage conversion coefficient is a custom coefficient that converts effective impact energy density into dimensionless damage. Its dimension is square meters per joule. Its function is to connect the energy input in the physical world with the damage measurement inside the model. Its value is obtained by regression analysis and calibration on known impact energy and the actual damage results.

[0053] Progressive damage increment The purpose is to simulate the fatigue propagation effect of existing micro-damage under continuous daily environmental stress. In this embodiment, the determination method refers to the material fatigue crack propagation theory and introduces the damage acceleration effect. The calculation formula is as follows:

[0054] ;

[0055] in: It represents the maximum temperature difference of the day, i.e. the temperature cycle range, in degrees Celsius (°C) or Kelvin (K). Its value is collected by the physical state sensing and event input module through the temperature sensor. This represents the average UV index for the day. It is dimensionless and its value is obtained by the physical state sensing and event input module through the meteorological data interface. , These are the temperature cycling damage coefficient and the ultraviolet damage coefficient, respectively, with dimensions of ℃⁻¹ and dimensionless, respectively. Both are custom model parameters, and their values ​​need to be calibrated through long-term outdoor exposure experiments or by fitting a large amount of historical operating data. Damage Acceleration Factor (TAF) is a mathematical expression used to characterize the accelerating effect of pre-existing potential damage on subsequent daily aging processes. Its purpose is to make the model closer to physical reality, that is, components with existing microcracks will age faster under the same environmental stress. It is based on the potential damage level at the previous time step. Obtained directly by calculation;

[0056] The potential damage level at the current moment is calculated. Subsequently, based on this result, the physical failure prediction model module needs to transform it into a failure probability map useful to the economic decision-making module, mapping an abstract physical state quantity to a risk index that conforms to the axioms of probability theory and ranges between 0 and 1. In this embodiment, the transformation process applies a preset S-shaped function and a preset critical damage threshold, as shown in the formula:

[0057] ;

[0058] in: The critical damage threshold is a core custom parameter. Its physical meaning is that when the potential damage of a component reaches this value, the probability of macroscopic failure is 50%. It represents the damage limit that this type of component can withstand. Its setting is based on the test of accelerated aging or mechanical loading to the point of failure of the same type of component. The steepness coefficient of the probability curve is a dimensionless parameter that determines how drastically the failure probability changes with increasing damage level; it is derived from statistical fitting based on historical failure data.

[0059] By specifying the physical failure prediction model module as described above, this invention achieves refined and mechanistic modeling of the damage process of photovoltaic equipment. It not only distinguishes between two core damage sources, instantaneous impact and daily aging, but also simulates the nonlinear characteristics of damage accumulation by introducing a damage acceleration factor. Furthermore, by setting an S-shaped function and a critical damage threshold, an abstract concept of damage degree is transformed into a clear and decision-making probability of future failure. Compared with simply proposing a prediction model in a general way, the technical solution of this embodiment has stronger physical realism, higher prediction accuracy, and better engineering feasibility.

[0060] Example 3

[0061] The economic scheduling and operation and maintenance decision module is also used to: determine the direct costs incurred by executing operation and maintenance decisions, and determine the additional expected risk costs caused by not executing operation and maintenance decisions; combine the direct costs and the additional expected risk costs, and after time value discounting, construct the objective function of the operation and maintenance scheduling model that minimizes risk costs;

[0062] The determination of the additional expected risk cost is based on the calculation of the probability of new failures within the current time interval and the total economic loss caused by a single failure; whereby the probability of new failures is derived from the time change of the probability of future failures in the failure probability map.

[0063] This embodiment is a further explanation of Embodiment 1, specifying the concrete implementation of the economic scheduling and operation and maintenance decision-making module. This module solves for the optimal strategy by constructing an objective function with the goal of minimizing risk costs. The core of this objective function is to weigh the direct costs of executing operation and maintenance decisions against the additional expected risk costs caused by not executing operation and maintenance decisions, and to discount the future costs using the time value of money. The determination of the additional expected risk costs is based on the probability of new failures within the current time interval.

[0064] In this embodiment, the core of the economic scheduling and operation and maintenance decision-making module is to construct and solve a risk-cost minimization operation and maintenance scheduling model; the objective function of this model aims to minimize the risk cost throughout the planning cycle. Within, the generalized risk cost over the entire lifecycle after discounting the time value. Its construction method is as follows:

[0065] ;

[0066] in: These are binary decision variables and are the core output of the model solution. Their function is to explicitly provide the decision variables for the first variable. The device is The operation and maintenance instructions at any time are data types of dimensionless binary variables, where 1 represents the decision to execute operation and maintenance, and 0 represents not to execute. The time discount factor is a coefficient used to discount future costs to their present value; its purpose is to enable models to prioritize recent high-risk events when making intertemporal decisions; the data type is dimensionless, and its value range is [range missing]. Its value can be set based on the company's benchmark rate of return or industry-standard rates of return. The total number of photovoltaic devices included in the management scope; The duration of the entire planning cycle;

[0067] The core of this objective function lies in the trade-off between two costs:

[0068] direct costs This refers to, if the decision variables This refers to the deterministic, one-time expenses incurred during operation and maintenance, which include spare parts procurement costs, labor costs of maintenance personnel, and power generation losses caused by equipment downtime during operation and maintenance.

[0069] Additional expected risk cost This refers to, if the decision variables This means that maintenance is not performed, resulting in the current time interval... Within this timeframe, the system needs to bear additional, uncertain potential losses. These losses are determined by multiplying the probability of new failures within the current time interval by the total economic loss caused by a single failure. The calculation formula is as follows:

[0070] ;

[0071] in: This represents the total economic loss caused by a single failure, expressed in yuan. It is a comprehensive cost, and its calculation needs to take into account the higher maintenance costs after the failure, possible safety accident compensation, and all power generation losses during the period from the occurrence of the failure to the completion of the repair. The probability of adding a fault is based on the known history of all past operational decisions. Under the conditions, among which Less than the current time At any point in the past, the device In time interval The conditional probability of the first failure within a given period serves to accurately measure the increase in risk in the current period resulting from delayed decision-making, thereby avoiding repeated calculations of future risks. Its source is the time change in the cumulative failure probability between two adjacent time points in the failure probability map output by the physical failure prediction model module, i.e.:

[0072] ;

[0073] By specifying the economic scheduling and operation and maintenance decision-making modules as described above, this invention constructs an objective function that balances current costs with future risks. It transforms the core of operation and maintenance decision-making from simply comparing current maintenance costs with a vague future risk to precisely balancing current maintenance costs with the expected additional losses incurred during the current period due to non-maintenance. By introducing the concepts of additional expected risk cost and time discount factor, the decision-making process can quantify the economic value of short-term intervention in reducing long-term total costs. This risk-averse value-based decision-making mechanism ensures that, with limited resources, the system's operation and maintenance decisions aim to maximize the reduction of total risk costs throughout the power plant's lifecycle.

[0074] Example 4

[0075] The system execution and feedback module is also used to: update the current potential damage level in response to the execution of maintenance work orders;

[0076] When the maintenance work order type is "replacement", the current potential damage level will be reset to the preset factory initial value.

[0077] When the type of maintenance work order is repair, the current potential damage level will be reset to the determined value based on the repair effect evaluation.

[0078] This embodiment further defines how the system execution and feedback module interacts in a closed loop with the physical failure prediction model module; specifically, the system execution and feedback module will respond to the execution of the maintenance work order and actively update the current potential damage level in the physical model. The update rules vary depending on the specific type of the maintenance work order.

[0079] In this embodiment, to achieve a dynamic closed loop between the physical model and economic decisions, the system execution and feedback module acts as a state updater; when the economic scheduling and operation and maintenance decision module outputs the optimal decision... That is, in Components at all times Once the maintenance operation is executed and the system execution and feedback module confirms the completion of the maintenance work order, the module will immediately trigger an action on the corresponding component in the physical failure prediction model module. The instruction to update the status parameters;

[0080] Specifically, the module will update the component at the next moment. Current potential damage level The purpose of this reset operation is to ensure that the physical model accurately reflects the real improvement effect of maintenance intervention on the physical state of the equipment. In this embodiment, the update rule for the damage level is defined as follows:

[0081] When the maintenance work order type is "replacement," it means that a brand new, intact component has replaced the old one. In this case, the system execution and feedback module will assess the potential damage to the component. Reset to a preset factory default value;

[0082] Initial factory value refers to a very small, non-zero value used to characterize the initial microscopic damage that may exist in a brand-new component during manufacturing and transportation; its source can be provided by the equipment manufacturer or set as a standard empirical value; for example... Will be reset to ,in It is usually close to 0;

[0083] When the maintenance work order type is "repair," such as repairing the junction box or bypass diode of a component, it indicates that the entire component has not been completely replaced. Its physical condition has been improved rather than fully restored to new condition. In this case, the system execution and feedback module will assess the potential damage to the component. Reset to a defined value based on the repair effectiveness assessment;

[0084] The definitive value for evaluating repair effectiveness refers to a value between 0 and the degree of damage before repair. Its purpose is to quantify the extent to which the repair operation restores the damage. A fixed reduction factor can be set based on the standard repair rate of the repair technique, for example... Alternatively, a more accurate damage value can be reassessed based on the results of non-destructive testing after repair.

[0085] By introducing this dynamic damage update mechanism in response to maintenance work orders, this invention establishes an effective feedback loop between physical prediction and economic decision-making, ensuring that the output of economic decisions accurately reflects changes in the physical model input. This enables the entire system to be adaptive: after performing maintenance, the system can immediately re-evaluate the future failure probability evolution based on the improved physical state and formulate a more accurate maintenance strategy for the next stage. This mechanism avoids the problem of the model still making incorrect predictions based on the old damage trajectory after a maintenance has been performed, thus improving the accuracy and effectiveness of the system's full lifecycle maintenance management.

[0086] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1.A photovoltaic device full life cycle operation and maintenance management system, characterized in that, The system comprises: a physical state sensing and event input module configured to collect physical parameters of external events associated with a photovoltaic device and to obtain real-time state data of the photovoltaic device; a physical failure prediction model module configured to deduce a damage accumulation process of the photovoltaic device based on the physical parameters and the real-time state data collected by the physical state sensing and event input module, and to calculate a failure probability graph representing a future failure probability of each device; an economic dispatching and operation decision module configured to receive the failure probability graph calculated by the physical failure prediction model module, to solve an optimal predictive maintenance strategy by taking minimizing a generalized risk cost of the photovoltaic device in a whole life cycle as an objective and by comprehensively considering preset economic constraints; a system execution and feedback module configured to generate an operation work order based on the optimal predictive maintenance strategy solved by the economic dispatching and operation decision module, to collect actual results after operation execution, and to use the actual results to feedback and calibrate the physical failure prediction model module; the economic dispatching and operation decision module is further configured to determine a direct cost generated by executing an operation decision and to determine a newly added expected risk cost caused by not executing the operation decision, to combine the direct cost and the newly added expected risk cost, and to construct an objective function of a generalized risk cost minimization operation dispatching model of the photovoltaic device in the whole life cycle by time value discounting; the newly added expected risk cost is determined based on a newly added failure probability in a current time interval and total economic losses caused by a single failure; the newly added failure probability is derived from a time variation of a future failure probability in the failure probability graph. 2.The photovoltaic device full life cycle operation and maintenance management system according to claim 1, characterized in that, the physical failure prediction model module is further configured to determine an instantaneous damage increment caused by the external events and to determine a gradual damage increment caused by daily environmental stress; by combining the instantaneous damage increment and the gradual damage increment, a current potential damage degree representing a micro-damage degree inside the device is iteratively updated based on a damage degree at a previous time. 3.The photovoltaic device full life cycle operation and maintenance management system according to claim 2, characterized in that, the instantaneous damage increment is determined based on impact energy and impact incidence angle of the external events, and by combining an impact area of the device and a preset material vulnerability coefficient representing a physical impact resistance capability of the device. 4.The photovoltaic device full life cycle operation and maintenance management system according to claim 2, characterized in that, the gradual damage increment is determined based on a temperature cycle range and a daily average ultraviolet index collected on the same day, and by applying a damage acceleration factor; the damage acceleration factor is used to represent an acceleration effect on a subsequent daily aging process determined by the current potential damage degree. 5.The photovoltaic device full life cycle operation and maintenance management system according to claim 2, characterized in that, the physical failure prediction model module is further configured to use the current potential damage degree as an input, to apply a preset S-shaped function for conversion, and to generate the failure probability graph; the conversion process of the S-shaped function applies a preset critical damage threshold. 6.The photovoltaic device full life cycle operation and maintenance management system according to claim 2, characterized in that, the system execution and feedback module is further configured to update the current potential damage degree in response to execution of the operation work order; when a type of the operation work order is replacement, the current potential damage degree is reset to a preset initial value at factory delivery. When the type of the operation and maintenance work order is maintenance, the current potential damage degree is reset to a determined value according to a maintenance effect evaluation.

Images