Moisture-proof control system for virtual power plant control

Abstract

The invention discloses a moisture-proof control system for virtual power plant control. The system comprises a humidity detection module, a control module and a moisture-proof execution module. The humidity detection module is used for monitoring real-time environment humidity data of to-be-protected areas of one or more distributed energy devices in the virtual power plant. The control module is arranged locally or on the edge side of the distributed energy equipment and comprises a communication unit, a data fusion unit and an intelligent decision-making unit. And the data fusion unit receives the humidity data and the multi-source management and control information from the central controller, and generates a data set through fusion. And the intelligent decision-making unit predicts the risk of humidity-related faults of the equipment under the current and future environmental conditions by using a multi-dimensional dynamic self-learning risk assessment model based on the data set, and selects a proper moisture-proof strategy. And the moistureproof execution module executes moistureproof operation according to the moistureproof control instruction. The system can effectively predict, prevent and control humidity-related faults in the virtual power plant, and the reliability and stability of the system are improved.

Description

Technical Field

[0001] This invention relates to virtual power plant control technology, and more particularly to a moisture-proof control system for virtual power plant control. This system is mainly used for moisture protection of distributed energy equipment. It optimizes moisture-proof strategies through real-time monitoring and intelligent decision-making, improving equipment reliability and stability and preventing humidity-related failures. Background Technology

[0002] With the rapid development of new energy technologies, Virtual Power Plants (VPPs), as an intelligent power dispatching model, are gradually becoming an important component of modern power systems. VPPs achieve efficient utilization of power resources by centrally managing and dispatching multiple distributed energy devices (such as wind power, solar power, and energy storage devices). As distributed energy devices become more widespread, their environmental adaptability requirements are also increasing. Especially given the impact of environmental factors such as humidity on equipment reliability, effectively controlling the damage caused by humidity has become an urgent problem to be solved.

[0003] Existing virtual power plant control systems typically monitor and manage distributed energy equipment through a centralized control platform. However, in practical applications, many systems do not fully consider the impact of humidity, an external environmental factor, on equipment operation. Both excessively high and low humidity levels can lead to equipment failure or performance degradation, especially in energy storage systems, power transmission equipment, and other sensitive electronic devices. While some existing technologies have attempted to protect equipment through environmental monitoring and temperature and humidity control, these are mostly simple humidity detection or rule-based control strategies, lacking dynamic assessment and intelligent protection strategies for humidity-related faults under complex environmental conditions.

[0004] In the process of developing this invention, the inventors discovered that existing humidity protection technologies often suffer from the following problems: First, traditional humidity monitoring technologies rely heavily on single environmental data, failing to fully integrate multiple influencing factors, such as equipment operating status, health status, and external environmental forecasts, resulting in insufficient accuracy and real-time performance of humidity protection strategies. Second, existing technologies are mostly fixed moisture control strategies, unable to dynamically adjust based on actual operating conditions and future forecast environmental conditions, leading to delayed or overly conservative system responses to humidity changes. Finally, most existing systems lack intelligent risk assessment models, failing to accurately predict the probability of humidity-related equipment failures, thus affecting the overall moisture protection effect and operating efficiency of the equipment. Therefore, a more intelligent, flexible, and comprehensive moisture control system is urgently needed to address the risk of equipment failure caused by humidity under complex environmental conditions. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a moisture-proof control system for virtual power plant control to solve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides a moisture-proof control system for virtual power plant control, wherein the virtual power plant includes a central controller and multiple distributed energy devices, and the system includes: A humidity detection module is installed in the protected area of ​​one or more distributed energy devices to monitor the real-time environmental humidity data of the protected area. A control module, located on the local or edge side of the distributed energy device, includes: A communication unit is used to establish a connection with the central controller, the humidity detection module, and the moisture-proof execution module. The data fusion unit is used to receive the real-time ambient humidity data; receive multi-source control information from the central controller; the multi-source control information includes: the operating status parameters of the distributed energy equipment; the real-time health status indicators of the distributed energy equipment; and the external environmental forecast information of the deployment site of the distributed energy equipment; and fuse the real-time ambient humidity data and the multi-source control information to generate a fused dataset. The intelligent decision-making unit is used to predict the probability of humidity-related failures of the distributed energy equipment under current and future forecast environmental conditions based on the fused dataset and a built-in risk assessment model based on multi-dimensional dynamic self-learning; based on the risk probability and the scheduling objectives and operating constraints issued by the central controller, it selects a moisture-proof strategy from the built-in moisture-proof strategy library; and generates moisture-proof control instructions according to the moisture-proof strategy. A moisture-proof execution module is installed in the area to be protected and is used to perform corresponding moisture-proof operations according to the moisture-proof control command.

[0007] The above technical solution has the following beneficial technical effects: This invention relates to a moisture-proof control system for virtual power plant control. By precisely deploying humidity detection modules in the protected areas of distributed energy devices, it can capture real-time changes in local environmental humidity, avoiding the problem of missing local humidity anomalies due to overall environmental monitoring. The control module is deployed locally or at the edge, reducing data transmission latency with the central controller, improving the response speed of moisture-proof control, and reducing the computational load on the central controller. The data fusion unit integrates real-time environmental humidity data with the operating status parameters, real-time health indicators, and external environmental forecast information of the deployment site, providing comprehensive and multi-dimensional data support for subsequent risk assessment. This avoids assessment bias caused by a single data dimension. The intelligent decision-making unit relies on a multi-dimensional dynamic self-learning risk assessment model to accurately predict the probability of humidity-related failures of equipment in the current and future environments. It also selects a moisture-proof strategy in conjunction with the scheduling objectives and operational constraints of the central controller. This effectively prevents humidity failures and avoids conflicts between moisture-proof operations and normal equipment operation requirements, ensuring the scheduling stability of the virtual power plant. Furthermore, the moisture-proof execution module accurately executes moisture-proof operations, improving the operational stability of distributed energy equipment in humid environments, reducing the number of equipment downtimes caused by humidity issues, and thus enhancing the operational reliability and economic benefits of the entire virtual power plant. Attached Figure Description

[0008] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein: Figure 1 This is a logical functional block diagram of a moisture-proof control system for virtual power plant control according to an embodiment of the present invention; Figure 2 This is a logical functional block diagram of the data fusion unit in an embodiment of the present invention; Figure 3 This is a logical functional block diagram of the intelligent decision-making unit according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the logical structure of the risk assessment model according to an embodiment of the present invention; Figure 5 This is a flowchart of the working process of the moisture-proof strategy selection subunit in an embodiment of the present invention; Figure 6 This is a flowchart of a moisture-proof control method executed by a moisture-proof control system according to an embodiment of the present invention; Figure 7 This is a flowchart of a moisture-proof control method executed by a control module according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of a computer system according to an embodiment of the present invention. Detailed Implementation

[0009] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0010] Example 1 Figure 1 This is a logical functional block diagram of a moisture-proof control system for virtual power plant control according to an embodiment of the present invention. Figure 1 As shown, this embodiment of the invention provides a moisture-proof control system for virtual power plant control. The virtual power plant includes a central controller and multiple distributed energy devices. The system includes: A humidity detection module is installed in the protected area of ​​one or more distributed energy devices to monitor the real-time environmental humidity data of the protected area. A control module, located on the local or edge side of the distributed energy device, includes: A communication unit is used to establish a connection with the central controller, the humidity detection module, and the moisture-proof execution module. The data fusion unit is used to receive the real-time ambient humidity data; receive multi-source control information from the central controller; the multi-source control information includes: the operating status parameters of the distributed energy equipment; the real-time health status indicators of the distributed energy equipment; and the external environmental forecast information of the deployment site of the distributed energy equipment; and fuse the real-time ambient humidity data and the multi-source control information to generate a fused dataset. The intelligent decision-making unit is used to predict the probability of humidity-related failures of the distributed energy equipment under current and future forecast environmental conditions based on the fused dataset and a built-in risk assessment model based on multi-dimensional dynamic self-learning; based on the risk probability and the scheduling objectives and operating constraints issued by the central controller, it selects a moisture-proof strategy from the built-in moisture-proof strategy library; and generates moisture-proof control instructions according to the moisture-proof strategy. A moisture-proof execution module is installed in the area to be protected and is used to perform corresponding moisture-proof operations according to the moisture-proof control command.

[0011] In a further embodiment, the moisture-proof strategy library contains multiple moisture-proof strategies, each of which is matched with different risk probability levels, the operating status of distributed energy devices, and the scheduling objectives of virtual power plants; the moisture-proof operation includes at least one of the following: starting the heater, starting the ventilation device, starting the dehumidifier, and adjusting the operating mode of the distributed energy device to generate internal heat; the area to be protected for the distributed energy device includes at least one of the following: the inside of the energy storage battery cabinet, the inside of the photovoltaic inverter cabinet, and the inside of the controller chassis.

[0012] The following example illustrates this in detail: This embodiment discloses a moisture-proof control system for virtual power plant control, which is applicable to a virtual power plant scenario that includes one central controller and 10 sets of distributed energy devices. The distributed energy devices include five sets of 10kW photovoltaic inverters and five sets of 50kWh energy storage battery packs. The central controller is deployed in the server cluster of the virtual power plant operation and maintenance center for unified management and control of the operation of all distributed energy devices.

[0013] The humidity detection module uses high-precision temperature and humidity sensors, with a total of 15 sensors deployed. Two sensors are installed on the inner wall of the heat dissipation cavity of each photovoltaic inverter, and one sensor is installed on the inner wall of the battery compartment of each energy storage battery pack. All sensors are fixed to the metal bracket in the area to be protected by clips, maintaining a safe distance of more than 5cm from the internal circuitry of the equipment. The module collects real-time environmental humidity data of the area to be protected every 10 seconds, with an accuracy controlled within ±2%RH and a collection range covering 0-100%RH, ensuring accurate capture of local humidity changes.

[0014] The control module uses a microcontroller as its core hardware. The control modules for the five photovoltaic inverters are directly integrated into the local control box of the equipment, while the control modules for the five energy storage battery packs are deployed in an edge computing gateway within a 100-meter radius of the equipment. The edge computing gateway maintains a communication connection with the energy storage battery packs via PoE power supply.

[0015] The communication unit adopts a dual-mode communication scheme of 4G / 5G and LoRa. The 4G / 5G module is used to establish a connection with the central controller, and the data transmission protocol is MQTT. The LoRa module is used to establish a connection with the humidity detection module and the moisture-proof execution module. The LoRa communication frequency is 433MHz and the transmission rate is 9.6kbps. Each humidity detection module is assigned a unique device address, and each moisture-proof execution module corresponds to a unique control interface to ensure the accuracy and uniqueness of data transmission.

[0016] The data fusion unit receives real-time environmental humidity data uploaded by the humidity detection module through the communication unit, and simultaneously receives multi-source control information from the central controller. The operating status parameters of the distributed energy equipment include the real-time output power, input voltage, and operating temperature of the photovoltaic inverter, as well as the remaining charge (SOC, %) and charging / discharging current of the energy storage battery pack. The real-time health status indicators of the distributed energy equipment include the aging degree of the internal capacitors of the photovoltaic inverter (assessed through capacitance decay rate, in %), the operating time of the cooling fan, and the number of charge / discharge cycles and individual battery voltage deviation (in mV) of the energy storage battery pack. External environmental forecast information for the deployment site of the distributed energy equipment includes hourly humidity, rainfall probability, and wind speed for the next 24 hours provided by the local meteorological department. The data fusion unit uses a weighted average algorithm to calibrate the real-time environmental humidity data and multi-source control information, with the weights for real-time environmental humidity data (0.4), operating status parameters (0.2), real-time health status indicators (0.2), and external environmental forecast information (0.2). Finally, a fused dataset containing timestamps, device numbers, and fused humidity-related parameters is generated and stored in JSON format.

[0017] The intelligent decision-making unit incorporates a multi-dimensional dynamic self-learning risk assessment model. The input dimensions of this model include real-time humidity values ​​from the fused dataset, cumulative equipment runtime, health status score (derived from health status indicators, with a maximum score of 100), and the predicted humidity value for the next 6 hours. This risk assessment model achieves self-learning through a gradient descent algorithm, updating its parameters every 7 days based on historical fault data and moisture-proofing effectiveness data to ensure prediction accuracy. The intelligent decision-making unit calculates the probability of humidity-related faults occurring in distributed energy equipment based on this risk assessment model. A probability ≤ 30% is classified as low risk, 30% < probability ≤ 60% as medium risk, and probability > 60% as high risk. Simultaneously, the intelligent decision-making unit receives scheduling targets (e.g., photovoltaic) from the central controller. The inverter must maintain 80% of its rated power output, and the energy storage battery pack must maintain ≥50% SOC during peak electricity consumption periods. Operating constraints (e.g., the photovoltaic inverter operating temperature must not exceed 60℃, and the energy storage battery pack charging / discharging current must not exceed 10A) are also considered. A matching strategy is selected from the built-in moisture-proof strategy library. This library includes three strategies: ventilation and heat dissipation for low risk, dehumidification and ventilation for medium risk, and dehumidification, heating, and sealing for high risk. The intelligent decision-making unit generates moisture-proof control instructions based on the selected strategy. The instruction format follows the Modbus-RTU protocol and includes the executing device number, operation type (e.g., start / stop dehumidifier, adjust heating element temperature, control ventilation fan speed), operation parameters (e.g., dehumidifier operating power, heating element target temperature, ventilation fan speed), and execution duration.

[0018] The moisture-proof execution module and humidity detection module are installed one-to-one in the area to be protected. The moisture-proof execution module for the photovoltaic inverter includes a small rotary dehumidifier (rated power 300W, dehumidification capacity 0.5L / h) and an axial flow fan (air volume 100m³ / h). The moisture-proof execution module for the energy storage battery pack includes a silicone heating pad (rated power 50W, heating temperature range 15-40℃) and a centrifugal ventilation fan (air volume 80m³ / h). The moisture-proof execution module receives moisture-proof control commands from the intelligent decision-making unit through the communication unit. When receiving dehumidification and ventilation strategy commands, the dehumidifier starts and runs at rated power, and the ventilation fan runs at 80% speed until the humidity in the area to be protected drops below 50%RH or the execution time set by the command is reached. When receiving dehumidification, heating and sealing strategy commands, the dehumidifier and heating pad start simultaneously. The heating pad maintains the temperature of the area to be protected at 20-25℃, and the ventilation vents of the equipment are closed until the risk probability drops below medium risk.

[0019] This invention's moisture-proof control system improves the humidity protection effect and operational reliability of distributed energy equipment in a virtual power plant. On one hand, the humidity detection module is precisely deployed in the protected area, enabling real-time detection of local humidity anomalies and avoiding blind spots caused by overall environmental monitoring. The control module is deployed locally or at the edge, reducing data transmission latency with the central controller to improve moisture-proof response speed and lowering the central controller's computational load. On the other hand, the data fusion unit integrates real-time humidity data with multi-source information such as equipment operating status, health indicators, and external environmental forecasts, providing comprehensive data support for risk assessment and avoiding judgment bias caused by single-dimensional data. The intelligent decision-making unit, relying on a multi-dimensional dynamic self-learning model, can accurately predict current and future humidity-related fault risks and select moisture-proof strategies based on the central controller's scheduling objectives and operational constraints. This effectively prevents humidity-related faults while avoiding conflicts between moisture-proof operations and normal equipment operation requirements. Finally, the moisture-proof execution module precisely executes the strategy, reducing the number of equipment downtimes due to humidity issues and significantly enhancing the overall operational stability and economic benefits of the virtual power plant.

[0020] Example 2 Figure 2 This is a logical functional block diagram of the data fusion unit according to an embodiment of the present invention. For example... Figure 2 As shown, the data fusion unit specifically includes: The data alignment subunit is used to establish a unified time axis, and to match and align the real-time environmental humidity data, the operating status parameters, the real-time health status indicators and the external environment forecast information at the corresponding time points on the time axis to form a time-synchronized multi-source data sequence. The feature vector construction subunit is used to extract the following feature values ​​for each processing time point on the time axis to construct a multi-dimensional feature vector: the environmental humidity sequence within the current and past predefined time windows; the operating status parameters and equipment operating mode identifiers at the current time point; the quantified values ​​of the real-time health status indicators at the current time point; and the predicted environmental humidity sequence and corresponding predicted equipment operating mode identifiers within the future predefined time windows. The dataset generation sub-unit is used to arrange the multi-dimensional feature vectors corresponding to multiple processing time points within a continuous time period in chronological order along the time axis to form the fused dataset, which is then used by the risk assessment model for time series analysis and prediction.

[0021] In some embodiments, the operating status parameters include at least one of the following: charging and discharging power, operating temperature, and grid connection status; the real-time health status indicators are calculated based on the real-time operating parameters of the distributed energy equipment monitored by multiple sensors and a predetermined health assessment model; the external environment forecast information is meteorological data for a future specified period of time at the deployment location of the distributed energy equipment, obtained by the virtual power plant central controller from an external meteorological service interface, including temperature, humidity, and precipitation probability.

[0022] In some embodiments, the real-time health status indicators are calculated through the following specific process: Multiple types of sensors deployed at key locations of the distributed energy equipment, including temperature sensors, vibration sensors, current and voltage sensors, and insulation monitoring sensors, continuously collect real-time operating parameters reflecting the operating status of the distributed energy equipment. These parameters include, but are not limited to, power device junction temperature, chassis vibration amplitude, output current harmonic distortion rate, and DC-side insulation resistance to ground. These real-time operating parameters are input into a predetermined health assessment model. This model first compares the actual measured value of each parameter with its corresponding preset health benchmark range to calculate the instantaneous deviation of each parameter. Subsequently, the health assessment model combines the instantaneous deviation of each parameter with two aging correction factors—historical operating time and cumulative workload—obtained from the historical database of the distributed energy equipment, to form a comprehensive feature vector. This health assessment model uses a multi-feature fusion algorithm based on random forest or support vector machine to train and infer this comprehensive feature vector. This process is not a simple weighted average, but rather uses machine learning algorithms to automatically learn the complex nonlinear mapping relationship between each feature and the health status of the distributed energy equipment, and accordingly assigns importance weights to different features for deep fusion. Ultimately, the health assessment model outputs a quantitative health status index value, which is a value between 0 and 1. A value of 1 represents that the distributed energy device is in a completely healthy state, while a value of 0 represents that the distributed energy device has lost its operational capability. This achieves a precise and quantitative representation of the current health status of the distributed energy device. This calculation process is executed periodically, allowing the health status index to dynamically reflect the latest health status of the distributed energy device.

[0023] The advantage of this data fusion unit lies in its ability to integrate heterogeneous multi-source data (environmental humidity, equipment operating parameters, health status, and external forecasts) from different sources and of different types into a highly structured, precisely synchronized fusion dataset through strict time alignment, the construction of multi-dimensional feature vectors, and the generation of time-series datasets. This process not only solves the problem of inconsistency in time scales among multi-source data and ensures the uniformity of the data baseline, but more importantly, it provides the backend machine learning-based risk assessment model with complete time-series feature inputs that simultaneously include historical patterns, current status, and future trends, improving the accuracy and reliability of the model's time-series analysis and prediction.

[0024] Example 3 Figure 3 This is a logical functional block diagram of the intelligent decision-making unit according to an embodiment of the present invention. Figure 3 As shown, the intelligent decision-making unit includes: The risk probability prediction subunit is used to calculate and output the risk probability of the distributed energy equipment experiencing humidity-related failures under the current actual and future forecast environmental conditions, based on the fused dataset and the risk assessment model based on multi-dimensional dynamic self-learning. The moisture prevention strategy selection subunit is used to receive the risk probability output by the risk probability prediction subunit, and, in combination with the scheduling target and operation constraints issued by the central controller, match and select an appropriate moisture prevention strategy from the moisture prevention strategy library. The control instruction generation subunit is used to generate a moisture-proof control instruction containing operation type, execution intensity and triggering conditions according to the moisture-proof strategy selection subunit, and send it to the moisture-proof execution module.

[0025] This embodiment is based on the aforementioned virtual power plant scenario, which includes one central controller, five 10kW photovoltaic inverters, and five 50kWh energy storage battery packs. The intelligent decision-making unit is implemented by the microcontroller of the control module, and the model inference response time is ≤500ms, ensuring rapid output of decision results.

[0026] The risk probability prediction subunit's built-in multi-dimensional dynamic self-learning risk assessment model adopts an LSTM (Long Short-Term Memory) architecture. The model input is a 16-dimensional feature vector generated by the data fusion unit (including historical humidity sequences, equipment operating parameters, health indicators, and future forecast humidity). The model's self-learning process is achieved as follows: every 7 days, it automatically retrieves historical data from the past 30 days (including equipment humidity fault records, moisture-proof operation logs, and environmental humidity change curves), and uses the gradient descent algorithm to update the model weights and optimize prediction accuracy. When calculating the risk probability, the model first normalizes the input feature vector (mapping the values ​​to the 0-1 range), and then performs feature operations through 3 hidden layers (64 neurons per layer), finally outputting a risk probability value of 0-100%. For example, under the condition that the current humidity is 68%RH and the forecast humidity for the next hour is 72%RH, the output risk probability of photovoltaic inverter No. 1 is 45%, and a probability confidence level (e.g., 92%) is generated simultaneously for subsequent strategy selection reference.

[0027] The moisture-proof strategy selection subunit pre-stores a moisture-proof strategy library adapted to the virtual power plant equipment. This library contains three core strategies and their corresponding triggering conditions: low-risk strategies (risk probability ≤ 30%) focus on ventilation and heat dissipation, activating only the ventilation fan; medium-risk strategies (30% < risk probability ≤ 60%) focus on dehumidification and ventilation, simultaneously activating the dehumidifier and ventilation fan; and high-risk strategies (risk probability > 60%) focus on dehumidification, heating, and sealing, activating the dehumidifier and heating elements while closing the equipment vents. After receiving the risk probability, the moisture-proof strategy selection subunit first reads the scheduling objectives (e.g., the photovoltaic inverter must maintain ≥ 8kW output power and the energy storage battery pack must maintain ≥ 50% SOC) and operating constraints (e.g., photovoltaic inverter operating temperature ≤ 60℃ and energy storage battery pack charging / discharging current ≤ 10A) issued by the central controller, and then performs strategy adaptation verification. For example, if the output risk probability of the No. 5 energy storage battery pack is 55% (medium risk), and the scheduling target requires it to enter the discharge state after 10:00 (to avoid the dehumidifier consuming power and affecting the energy storage capacity), then the adjustment strategy is low-power dehumidification (200W) and ventilation to ensure that moisture prevention does not conflict with the scheduling target. Finally, the appropriate strategy is selected and the strategy number is marked.

[0028] The control command generation subunit generates standardized moisture-proof control commands based on the operational requirements of the selected strategy. These commands, using the Modbus-RTU protocol format, contain three parts: operation type (starting the dehumidifier, adjusting the ventilation fan speed, and starting the heating element); execution intensity (dehumidifier operating power 200W, ventilation fan speed 1200r / min, and heating element target temperature 25℃); and triggering conditions (stopping dehumidification when humidity drops below 55%RH and reassessing the risk after 30 minutes of continuous operation). The commands also carry a unique device identifier and a validity period. For example, if the unique device identifier is BAT-05, it corresponds to battery pack No. 5. After generation, the command is sent to the corresponding moisture-proof execution module via the LoRa communication unit of the control module. This LoRa communication unit operates at a frequency of 433MHz. If a communication interruption causes the command transmission to fail, a 4G / 5G backup communication channel is triggered for retransmission.

[0029] The beneficial technical effect of the intelligent decision-making unit lies in its ability to achieve closed-loop intelligent decision-making from data analysis to control execution through the collaborative work of three sub-units: risk probability prediction, intelligent strategy selection, and precise instruction generation. This design dynamically couples fault risk prediction based on multi-dimensional data with power grid dispatching objectives and equipment operating constraints. This enables it to proactively and adaptively select the optimal moisture-proof strategy while ensuring equipment safety, ultimately generating directly executable control instructions with clearly defined parameters.

[0030] Figure 4 This is a schematic diagram of the logical structure of the risk assessment model according to an embodiment of the present invention. Figure 4As shown, in a further embodiment, the risk assessment model includes: The feature mapping layer is used to map the real-time environmental humidity data, operating status parameters, real-time health status indicators and external environmental forecast information in the fused dataset into humidity stress features, load fluctuation features, aging sensitivity features and environmental trend features, respectively. The dynamic weight allocation unit is used to execute dynamic weight allocation rules based on the preset humidity sensitivity threshold, preset stable operating range, preset health benchmark value and preset environmental safety threshold of the distributed energy equipment; the dynamic weight allocation rules include: assigning dynamic weights to the humidity stress feature, the load fluctuation feature, the aging sensitivity feature and the environmental trend feature, and outputting the multi-dimensional fusion feature after dynamic weight weighting; The risk calculation unit is used to output the predicted risk probability of humidity-related faults within a preset time period based on the multi-dimensional features after dynamic weighting and the long short-term memory network time-series prediction model trained based on historical humidity fault sample data. The self-updating unit is used to periodically compare the actual operational fault data of the distributed energy equipment with the predicted risk probability, and correct the dynamic weight allocation rule and the network parameters of the long short-term memory network time series prediction model through error feedback, so as to realize the adaptive optimization of the risk assessment model.

[0031] In some embodiments, the mapping formula for humidity stress characteristics is as follows: ;or, ; Among them, H cur This is the current real-time ambient humidity value; H th It is a preset humidity safety threshold; H sat α is the effective upper limit of humidity; HS is the humidity stress characteristic value, which represents the degree of stress of environmental humidity on the equipment; α and β are model parameters preset according to the characteristics of the equipment; H(t) is the environmental humidity value at any time t; t0 and t are the time range of integration.

[0032] In some embodiments, the mapping formula for load fluctuation characteristics is as follows: ;or, ; Where P is the equipment operating power sequence; LF is the load fluctuation characteristic value, representing the intensity of load fluctuation; σ(P) is the standard deviation of the power sequence P; μ(P) is the average value of the power sequence P; ΔP is the power change threshold, used to determine whether drastic fluctuations have occurred; T is the time window length; P t and P t-1These are the power values ​​of the device at time points t and t-1; Count is the power value that satisfies condition |P t P t 1 |>ΔP| times.

[0033] In some embodiments, the mapping formula for aging-sensitive characteristics is as follows: ;or, ; Among them, S cur This is the current health status indicator value of the equipment; S new is the baseline value of the equipment's health status at the factory; AS is the aging-sensitive characteristic value, representing the degree of aging of the equipment; k and λ are model parameters preset according to the equipment characteristics, describing the changing pattern of aging.

[0034] In some embodiments, the mapping formula for environmental trend characteristics is as follows: ;or, ; Among them, H fcast It is the average humidity value predicted for the future; H cur τ is the current real-time ambient humidity value; τ is the length of the future time period used to calculate the rate of change of the trend; ET is the environmental trend characteristic value, representing the trend of humidity change; w i It is the preset weight of the i-th type of weather event; I i This indicates whether the forecast includes the i-th type of weather event (1 indicates inclusion, 0 indicates exclusion); ∑ is a weighted summation operation used to calculate the weighted trend characteristics.

[0035] This embodiment is based on the aforementioned virtual power plant scenario, which includes one central controller, five 10kW photovoltaic inverters, and five 50kWh energy storage battery packs. The risk assessment model relies on the microcontroller of the control module to perform inference operations. The model training phase is completed on the server of the central controller. The training dataset uses the historical operating data of the distributed energy equipment over the past year to ensure that the model can accurately adapt to the characteristics of the virtual power plant equipment.

[0036] The feature mapping layer uses a standardized mapping algorithm to convert the four types of input information in the fused dataset into features of a unified dimension: For real-time ambient humidity data, the deviation rate between the current humidity value and the device's preset humidity sensitivity threshold is first calculated, and then the deviation rate is mapped to a humidity stress feature in the 0-1 range. The larger the deviation rate, the closer the feature value is to 1. For operating status parameters, the ratio of the device's real-time output power to its rated power and the amplitude of input voltage fluctuation are extracted and mapped to a load fluctuation feature in the 0-1 range through weighted averaging. The larger the fluctuation, the closer the feature value is to 1. For real-time health status indicators, the quantified value of capacitor aging degree and the number of battery cycles are compared with preset health benchmark values, and the decay rate is calculated and mapped to an aging sensitivity feature in the 0-1 range. The more severe the decay, the closer the feature value is to 1. For external environment forecast information, the deviation ratio between the maximum forecast humidity value for the next hour and the preset environmental safety threshold is taken and mapped to an environmental trend feature in the 0-1 range. The larger the deviation, the closer the feature value is to 1. Finally, a feature vector composed of four standardized feature values ​​is output. Specifically, in the embodiment described, the formula for calculating the attenuation rate is: Attenuation rate = (Preset health baseline value - Current real-time quantized value) / Preset health baseline value. This formula objectively characterizes the aging degree of the device by calculating the relative loss ratio of the current health status quantized value (e.g., capacitor capacity or battery health) to its ideal initial state (i.e., the preset health baseline value).

[0037] The dynamic weight allocation unit pre-sets differentiated thresholds for different types of equipment: For photovoltaic inverters, the preset humidity sensitivity threshold is 65%RH, the preset stable operating range is 5-10kW output power, the preset health benchmark is 80 minutes of capacitor aging, and the preset environmental safety threshold is 70%RH predicted humidity; for energy storage battery packs, the preset humidity sensitivity threshold is 55%RH, the preset stable operating range is 30%-100% SOC, the preset health benchmark is 1000 battery cycles, and the preset environmental safety threshold is 60%RH predicted humidity. The dynamic weight allocation rule is that when a certain characteristic value is close to the corresponding preset threshold, the weight of that characteristic is increased; when it is far from the threshold, the weight is decreased. Taking photovoltaic inverter No. 1 as an example, its current humidity stress characteristic value is 0.8, load fluctuation characteristic value is 0.3, aging sensitivity characteristic value is 0.4, and environmental trend characteristic value is 0.5. The corresponding dynamic weights are 0.4, 0.2, 0.2, and 0.2, respectively. The weighted multi-dimensional fused characteristics are then calculated by multiplying the characteristic value by the corresponding weight.

[0038] The long short-term memory network time-series prediction model of the risk calculation unit adopts a three-layer network architecture. The input layer receives multi-dimensional fusion features weighted by dynamic weights and associates them with the fusion feature sequence of the past 5 minutes, with a total of 5 input nodes. The hidden layer contains two layers, each with 32 neurons, and the activation function is ReLU. The output layer has one neuron, and the activation function is Sigmoid, with the output range being 0-100% risk probability. During the model training phase, the Adam optimizer is used with a learning rate of 0.001 and the mean squared error loss function. It is iteratively trained for 500 epochs until the loss value stabilizes below 0.005. When making predictions, the model first inputs the fusion feature sequence of the current moment and the past 5 minutes. After processing by the long short-term memory network, it outputs the probability of humidity-related fault risk in the next hour, and also outputs the prediction confidence score for the intelligent decision-making unit to refer to.

[0039] The self-updating unit is set to execute the update process every Monday at 2:00 AM, a low-load period for distributed energy equipment, thus avoiding disruption to normal equipment operation during the update process. During the update, the system first retrieves the actual operating data of the equipment from the central controller over the past seven days, including whether humidity-related faults occurred, the ambient humidity at the time of the fault, and equipment operating parameters. This data is then compared with the predicted risk probability output by the model during the same period. The prediction error is calculated using the mean absolute error (MAE). If the error is less than 5%, only the neuron weights of the hidden layer of the Long Short-Term Memory (LSTM) network are fine-tuned. If the error exceeds 5%, the dynamic weight allocation rules are corrected first, and then the parameters of the LSM network output layer are retrained. After the update, the model accuracy is verified using historical fault samples to ensure that the prediction accuracy improvement after the update is no less than 2%. For example, after a certain update, the model's prediction accuracy for humidity faults in photovoltaic inverters improved from 91% to 93%, thus achieving adaptive optimization to adapt to dynamic scenarios such as equipment aging and environmental changes.

[0040] The advantages of this risk assessment model are as follows: the feature mapping layer standardizes multi-source information in the fused dataset into four types of features, including humidity stress and load fluctuation, solving the problem of inconsistent data dimensions; the dynamic weight allocation unit dynamically adjusts feature weights based on the preset threshold of the equipment, giving more importance to key features close to the risk threshold and avoiding assessment bias caused by fixed weights; the risk calculation unit, combined with the LSTM time series prediction model, can accurately output the risk probability for a preset time period based on historical fault data, avoiding the limitations of short-term judgments relying solely on current data; the self-updating unit corrects parameters by comparing actual fault data with prediction results, enabling the model to adapt to dynamic scenarios such as equipment aging and environmental changes, maintaining high prediction accuracy over the long term, and ultimately providing a reliable basis for selecting moisture-proof strategies, effectively reducing the probability of humidity-related faults in distributed energy equipment.

[0041] In a further embodiment, the dynamic weight allocation rule includes: The weight of the humidity stress feature increases in a stepwise manner as the real-time humidity value approaches the preset humidity sensitivity threshold, as shown in the following formula: ; Among them, W h H represents the humidity stress feature weights, where H is the real-time humidity value, T1 to T3 are preset humidity thresholds, and w1 to w4 are the corresponding weight values ​​that increase sequentially.

[0042] The weight of the aging-sensitive feature increases linearly with the proportion of real-time health status indicators deteriorating beyond a preset health benchmark value, and its calculation formula is as follows: Wa = k1× max(0, (B - H) / B); Where Wa is the weight of aging-sensitive features, H is the real-time health status index value, B is the preset health benchmark value, and k1 is the linear growth coefficient.

[0043] The weight of the load fluctuation characteristic increases with the proportion of the fluctuation amplitude of the operating state parameter exceeding the preset stable operating range, and its calculation formula is as follows: W l = k2× max(0, (F - S) / S); Among them, W l is the load fluctuation characteristic weight, F is the fluctuation amplitude of the operating status parameter, S is the preset stable operating range threshold, and k2 is the proportional coefficient.

[0044] The weight of the environmental trend feature increases with the proportion of the future humidity increase rate exceeding the preset environmental safety threshold in the external environmental forecast information, and its calculation formula is as follows: We = k3× max(0, (R - E) / E); Wherein, We is the environmental trend characteristic weight, R is the future rate of humidity increase, E is the preset environmental safety threshold, and k3 is the proportional coefficient. k1, k2, and k3 are coefficients determined through experimental data analysis, model optimization, or engineering experience, used to describe the system's response to humidity, health status, load status, and environmental changes.

[0045] Figure 5 This is a flowchart illustrating the working process of the moisture-proof strategy selection subunit in an embodiment of the present invention. Figure 5 As shown, in a further embodiment, the moisture-proof strategy selection subunit is specifically used to perform the following steps: S51: Based on the risk probability and the real-time health status indicators of the distributed energy equipment, determine the dynamic risk level and screen candidate moisture-proof strategies based on the dynamic risk level; S52: Eliminate candidate moisture-proof strategies that exceed the operational constraints to obtain the remaining candidate moisture-proof strategies; S53: Based on the scheduling target issued by the central controller, determine the final suitable moisture-proof strategy from the remaining candidate moisture-proof strategies.

[0046] This embodiment is based on the aforementioned virtual power plant scenario including one central controller, five 10kW photovoltaic inverters, and five 50kWh energy storage battery packs. The moisture protection strategy selection process for photovoltaic inverter No. 2 (equipment number PV-02) is used as an example for detailed explanation. The moisture protection strategy selection subunit first receives the current risk probability of PV-02 (42%) output by the risk probability prediction subunit. Simultaneously, it retrieves the real-time health status indicators of the device from the data fusion unit, including a capacitor aging quantification value of 75 points and a cooling fan operating time of 5000 hours. The device's preset health benchmark values ​​are a capacitor aging value of 80 points and a preset safe operating time for the cooling fan of 8000 hours. Combined with the preset dynamic risk level judgment rules, a low risk is defined as a risk probability not exceeding 30% and a health indicator not lower than 90% of the benchmark value; a risk probability exceeding 30% but not exceeding 60% or a health indicator below the benchmark value is defined as a low risk. Values ​​between 80% and 90% are considered medium risk, while a risk probability greater than 60% or health indicators below the baseline value of 80% are considered high risk. The dynamic risk level of PV-02 is determined to be medium risk, and candidate moisture-proof strategies corresponding to medium risk are selected from the moisture-proof strategy library. Specifically, these include standard dehumidification plus ventilation strategy, low-power dehumidification plus ventilation strategy, and intermittent dehumidification plus ventilation strategy. The standard dehumidification power is 300 watts and the ventilation fan speed is 100%, the low-power dehumidification power is 200 watts and the ventilation fan speed is 80%, and the intermittent dehumidification power is 300 watts, running for 10 minutes and stopping for 5 minutes, with the ventilation fan speed at 90%. Next, the moisture-proof strategy selection subunit reads the PV-02 operating constraints issued by the central controller, specifically: the operating temperature should not exceed 60 degrees Celsius, the maximum operating power of the dehumidifier should not exceed 300 watts, and the total power consumption of the equipment should not exceed 10.5 kilowatts to avoid affecting the output power. The candidate moisture-proof strategies are constrained and verified. Among them, when the standard dehumidification plus ventilation strategy is running, the total power consumption of the equipment is 10 kilowatts of output power plus 300 watts of dehumidification power plus 150 watts of ventilation power, totaling 10.45 kilowatts, which is close to the upper limit of total power consumption. Moreover, in a high-temperature environment, the operating temperature of the equipment will rise to 62 degrees Celsius, exceeding the operating temperature constraint. Therefore, this strategy is eliminated, leaving two candidate strategies: low-power dehumidification plus ventilation strategy and intermittent dehumidification plus ventilation strategy. Finally, the moisture-proof strategy selection subunit was combined with the scheduling target issued by the central controller, namely, that PV-02 needs to maintain a stable output power of no less than 9 kW during the peak power consumption period from 10:00 to 12:00, and the total power consumption is allowed not to exceed 10.3 kW. The scheduling adaptability analysis of the remaining two candidate strategies was carried out. When the low-power dehumidification plus ventilation strategy is running, the total power consumption of the equipment is 9 kW output power plus 200 kW dehumidification power plus 120 kW ventilation power, totaling 10.2 kW, which can stably maintain the 9 kW output power and meet the scheduling target. Although the intermittent dehumidification plus ventilation strategy has a lower average power consumption, the intermittent start and stop of the dehumidifier may cause local humidity fluctuations in the area to be protected by the equipment, affecting the stability of the equipment operation. Therefore, the low-power dehumidification plus ventilation strategy was finally determined to be the moisture-proof strategy suitable for PV-02.

[0047] In a further embodiment, the moisture-proof strategy selection subunit can also be used to: call the built-in strategy effect prediction model to predict the impact of the finally adapted moisture-proof strategy on the real-time output, operating energy consumption and health status degradation rate of the equipment, and generate strategy effect evaluation data; and feed back the strategy effect evaluation data corresponding to the finally adapted moisture-proof strategy to the risk assessment model.

[0048] In a further embodiment, determining the dynamic risk level based on the risk probability and the real-time health status indicators of the distributed energy equipment, and then screening candidate moisture-proof strategies based on the dynamic risk level, specifically includes: When the risk probability is lower than the preset low-risk threshold and the real-time health status indicators are better than the preset health threshold, it is determined to be low-risk, and an intermittent ventilation and moisture-proof strategy is adopted. When the risk probability is between the preset low risk threshold and the preset high risk threshold and the real-time health status indicator is within the preset health threshold range, it is determined to be medium risk, and a constant temperature dehumidification and moisture-proof strategy is adopted. When the risk probability is higher than the preset high-risk threshold or the real-time health status indicator is worse than the preset health threshold, it is determined to be high-risk, and a moisture-proof strategy of shutting down dehumidification and isolation protection is adopted. The operational constraints include any number of the following: the rated power range of the equipment, the additional energy consumption limit of the virtual power plant, and the continuous operating time limit of the equipment.

[0049] In a further embodiment, determining the final suitable moisture-proof strategy from the remaining candidate moisture-proof strategies based on the scheduling target issued by the central controller specifically includes: Analyze the scheduling objective to obtain one or more optimization sub-objectives and their priority weights; For each remaining candidate moisture-proof strategy, predict its impact on the one or more optimization sub-objectives after execution, and calculate a comprehensive score based on the priority weights; Compare the overall scores of all remaining candidate moisture-proof strategies, and determine the candidate moisture-proof strategy with the highest overall score as the final suitable moisture-proof strategy.

[0050] In a further embodiment, the specific process by which the moisture-proof strategy selection subunit determines the final suitable moisture-proof strategy from the remaining candidate moisture-proof strategies based on the scheduling objective issued by the central controller is as follows: First, the moisture-proof strategy selection subunit parses the received scheduling objective, decomposes it into multiple quantifiable optimization sub-objectives, and assigns a dynamic priority weight to each optimization sub-objective; for example, optimization sub-objectives may include minimizing operating costs per unit time, maximizing power supply reliability, or maximizing overall energy efficiency, and their weights are dynamically adjusted according to the current overall operating conditions of the virtual power plant. Subsequently, for each remaining candidate moisture-proof strategy, the moisture-proof strategy selection subunit calls the built-in performance prediction model, and based on the historical data and state prediction algorithm of the distributed energy equipment, quantifies and predicts the specific impact values ​​of executing the remaining candidate moisture-proof strategy on the above-mentioned optimization sub-objectives, such as the predicted additional operating costs, the expected reduction in downtime due to failure, and the percentage improvement in the operating energy efficiency of the distributed energy equipment. Next, the moisture-proof strategy selection subunit converts each influence value into a dimensionless utility score using a pre-set normalization algorithm (e.g., min-max normalization). It then uses the priority weights to perform a weighted sum of these utility scores, thereby calculating the comprehensive score for each candidate strategy. Finally, the moisture-proof strategy selection subunit compares the comprehensive scores of all candidate strategies and determines the strategy with the highest score as the ultimately suitable moisture-proof strategy.

[0051] In a further embodiment, the moisture-proof strategy selection subunit quantifies and predicts the specific impact values ​​of candidate moisture-proof strategies on each optimization sub-objective through a built-in performance prediction model. The specific implementation process is as follows: First, the moisture-proof strategy selection subunit constructs a multi-dimensional input feature vector containing the operating status of the distributed energy equipment, environmental conditions, and moisture-proof strategy parameters. The operating status features of the distributed energy equipment include the current power output, temperature of the distributed energy equipment, and insulation resistance value. The environmental condition features include the real-time humidity value and humidity change trend. The moisture-proof strategy parameters are encoded according to the specific content of the candidate strategies. For example, the heating and dehumidification strategy is encoded as the power level and duration, and the ventilation strategy is encoded as the fan speed and operating cycle. Next, based on a Long Short-Term Memory (LSTM) network architecture, the performance prediction model simulates the state evolution of distributed energy devices after implementing candidate moisture-proof strategies by accessing similar operating condition data from the historical operating database of distributed energy devices. For the operating cost sub-objective, the model calculates the sum of additional energy consumption costs and maintenance costs during strategy execution. For the power supply reliability sub-objective, the model analyzes the health status degradation curves of distributed energy devices to predict the avoidable downtime and corresponding reduction in power supply loss under strategy protection. For the energy efficiency sub-objective, the model assesses the impact of strategy implementation on the operating energy efficiency of distributed energy devices and calculates the degree of output power improvement per unit of energy consumption. Finally, the performance prediction model outputs the quantified impact values ​​for each sub-objective. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application.

[0052] Figure 6 This is a flowchart of a moisture-proof control method executed by a moisture-proof control system according to an embodiment of the present invention. Figure 6 As shown, this embodiment of the invention also provides a computer-readable storage medium storing a computer program thereon. When executed by a processor, the program implements a moisture-proof control method for virtual power plant control. The virtual power plant includes a central controller and multiple distributed energy devices. The method includes the following steps: Step S61: Monitor the real-time environmental humidity data of the protected area by setting a humidity detection module in one or more distributed energy devices in the protected area; Step S62: Using the control module located locally or at the edge of the distributed energy device, perform the following operations: Establish a connection with the central controller, the humidity detection module, and the moisture-proof execution module using the communication unit of the control module; receive real-time environmental humidity data and multi-source control information from the central controller; the multi-source control information includes the operating status parameters of the distributed energy device, the real-time health status indicators of the distributed energy device, and external environmental forecast information of the deployment location of the distributed energy device; use the data fusion unit of the control module to fuse the real-time environmental humidity data and the multi-source control information to generate a fused dataset. Step S63: The intelligent decision-making unit of the control module performs the following operations: Based on the fused dataset, using a built-in risk assessment model based on multi-dimensional dynamic self-learning, predicts the probability of humidity-related failures of the distributed energy equipment under current and future forecast environmental conditions; based on the risk probability and the scheduling objectives and operational constraints issued by the central controller, selects a moisture-proof strategy from the built-in moisture-proof strategy library; and generates a moisture-proof control command according to the moisture-proof strategy. Step S64: The moisture-proof execution module set in the area to be protected receives the moisture-proof control command and performs the corresponding moisture-proof operation according to the moisture-proof control command.

[0053] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms.

[0054] Figure 7 This is a flowchart illustrating a method executed by a control module located locally or at the edge of a distributed energy device, according to an embodiment of the present invention. Figure 7As shown, the present invention also provides an electronic device comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a moisture-proof control method for virtual power plant control provided by the present invention, executed by a control module located locally or at the edge of a distributed energy device, the method comprising the following steps: S71: Receives real-time ambient humidity data of the area to be protected, monitored by the humidity detection module; S72: Receive multi-source control information from the central controller, the multi-source control information including the operating status parameters of the distributed energy equipment, real-time health status indicators and external environment forecast information of the deployment site; S73: Integrate the real-time environmental humidity data with the multi-source control information to generate a fused dataset; S74: Based on the fused dataset, using the built-in risk assessment model based on multi-dimensional dynamic self-learning, predict the probability of humidity-related failures of distributed energy equipment under current and future forecast environmental conditions; S75: Based on the risk probability and in conjunction with the scheduling objectives and operational constraints issued by the central controller, select an appropriate moisture-proof strategy from the built-in moisture-proof strategy library. S76: Based on the selected moisture-proof strategy, generate moisture-proof control instructions and send them to the moisture-proof execution module to drive it to perform the corresponding moisture-proof operation.

[0055] The following is for reference. Figure 8 This diagram illustrates a structural schematic of a computer system 800 suitable for implementing embodiments of the present invention in an electronic device. The computer system 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 802 or a program loaded from a storage portion 808 into a random access memory (RAM) 803. The RAM 803 also stores various programs and data required for the operation of the computer system 800. The CPU 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0056] The following components are connected to I / O interface 805: an input section 806 including a keyboard, mouse, etc.; an output section 807 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to I / O interface 805 as needed. A removable medium 811, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 810 as needed so that computer programs read from it can be installed into storage section 808 as needed.

[0057] In particular, according to the embodiments disclosed in this invention, the processes described in the above main step diagrams can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the main step diagrams. In the above embodiments, the computer program can be downloaded and installed from a network via communication section 809, and / or installed from removable medium 811. When the computer program is executed by central processing unit 801, it performs the functions defined in the system of this invention.

[0058] It should be noted that the computer-readable medium shown in this invention may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof.

[0059] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A moisture-proof control system for controlling a virtual power plant, the virtual power plant comprising a central controller and multiple distributed energy devices, characterized in that, The system includes: A humidity detection module is installed in the protected area of ​​one or more distributed energy devices to monitor the real-time environmental humidity data of the protected area. A control module, located on the local or edge side of the distributed energy device, includes: A communication unit is used to establish a connection with the central controller, the humidity detection module, and the moisture-proof execution module. The data fusion unit is used to receive the real-time ambient humidity data; receive multi-source control information from the central controller; the multi-source control information includes: the operating status parameters of the distributed energy equipment; the real-time health status indicators of the distributed energy equipment; and the external environmental forecast information of the deployment site of the distributed energy equipment; and fuse the real-time ambient humidity data and the multi-source control information to generate a fused dataset. The intelligent decision-making unit is used to predict the probability of humidity-related failures of the distributed energy equipment under current and future forecast environmental conditions based on the fused dataset and a built-in risk assessment model based on multi-dimensional dynamic self-learning; based on the risk probability and the scheduling objectives and operating constraints issued by the central controller, it selects a moisture-proof strategy from the built-in moisture-proof strategy library; and generates moisture-proof control instructions according to the moisture-proof strategy. A moisture-proof execution module is installed in the area to be protected and is used to perform corresponding moisture-proof operations according to the moisture-proof control command.

2. The system as described in claim 1, characterized in that, The intelligent decision-making unit includes: The risk probability prediction subunit is used to calculate and output the risk probability of the distributed energy equipment experiencing humidity-related failures under the current actual and future forecast environmental conditions, based on the fused dataset and the risk assessment model based on multi-dimensional dynamic self-learning. The moisture prevention strategy selection subunit is used to receive the risk probability output by the risk probability prediction subunit, and, in combination with the scheduling target and operation constraints issued by the central controller, match and select an appropriate moisture prevention strategy from the moisture prevention strategy library. The control instruction generation subunit is used to generate a moisture-proof control instruction containing operation type, execution intensity and triggering conditions according to the moisture-proof strategy selection subunit, and send it to the moisture-proof execution module.

3. The system as described in claim 1 or 2, characterized in that, The risk assessment model includes: The feature mapping layer is used to map the real-time environmental humidity data, operating status parameters, real-time health status indicators and external environmental forecast information in the fused dataset into humidity stress features, load fluctuation features, aging sensitivity features and environmental trend features, respectively. The dynamic weight allocation unit is used to execute dynamic weight allocation rules based on the preset humidity sensitivity threshold, preset stable operating range, preset health benchmark value and preset environmental safety threshold of the distributed energy equipment; the dynamic weight allocation rules include: assigning dynamic weights to the humidity stress feature, the load fluctuation feature, the aging sensitivity feature and the environmental trend feature, and outputting the multi-dimensional fusion feature after dynamic weight weighting; The risk calculation unit is used to output the predicted risk probability of humidity-related faults within a preset time period based on the multi-dimensional features after dynamic weighting and the long short-term memory network time-series prediction model trained based on historical humidity fault sample data. The self-updating unit is used to periodically compare the actual operational fault data of the distributed energy equipment with the predicted risk probability, and correct the dynamic weight allocation rule and the network parameters of the long short-term memory network time series prediction model through error feedback, so as to realize the adaptive optimization of the risk assessment model.

4. The system as described in claim 3, characterized in that, The dynamic weight allocation rules include: The weight of the humidity stress feature increases in a stepwise manner as the real-time humidity value approaches the preset humidity sensitivity threshold. The weight of the aging-sensitive feature increases linearly with the proportion of the real-time health status index deteriorating beyond the preset health benchmark value. The weight of the load fluctuation characteristic increases with the proportion of the fluctuation range of the operating status parameter exceeding the preset stable operating range. The weight of the environmental trend feature increases with the proportion of the rate of increase of future humidity in the external environmental forecast information that exceeds the preset environmental safety threshold.

5. The system as described in claim 3, characterized in that, The data fusion unit specifically includes: The data alignment subunit is used to establish a unified time axis, and to match and align the real-time environmental humidity data, the operating status parameters, the real-time health status indicators and the external environment forecast information at the corresponding time points on the time axis to form a time-synchronized multi-source data sequence. The feature vector construction subunit is used to extract the following feature values ​​for each processing time point on the time axis to construct a multi-dimensional feature vector: the environmental humidity sequence within the current and past predefined time windows; the operating status parameters and equipment operating mode identifiers at the current time point; the quantified values ​​of the real-time health status indicators at the current time point; and the predicted environmental humidity sequence and corresponding predicted equipment operating mode identifiers within the future predefined time windows. The dataset generation sub-unit is used to arrange the multi-dimensional feature vectors corresponding to multiple processing time points within a continuous time period in chronological order along the time axis to form the fused dataset, which is then used by the risk assessment model for time series analysis and prediction.

6. The system as described in claim 2, characterized in that, The moisture-proof strategy selection subunit is specifically used for: Based on the risk probability and the real-time health status indicators of the distributed energy equipment, a dynamic risk level is determined and candidate moisture-proof strategies are screened based on the dynamic risk level. Candidate moisture-proof strategies that exceed the operational constraints are eliminated to obtain the remaining candidate moisture-proof strategies; Based on the scheduling target issued by the central controller, the final suitable moisture-proof strategy is determined from the remaining candidate moisture-proof strategies.

7. The system as described in claim 6, characterized in that, The moisture-proof strategy selection subunit is also used for: The built-in strategy effect prediction model is invoked to predict the impact of the finally adapted moisture-proof strategy on the real-time output, operating energy consumption and health status degradation rate of the equipment, and generate strategy effect evaluation data. The strategy effectiveness evaluation data corresponding to the final adapted moisture-proof strategy is fed back to the risk assessment model.

8. The system as described in claim 6, characterized in that, The process of determining a dynamic risk level based on the risk probability and the real-time health status indicators of distributed energy devices, and then screening candidate moisture-proof strategies based on the dynamic risk level, specifically includes: When the risk probability is lower than the preset low-risk threshold and the real-time health status indicators are better than the preset health threshold, it is determined to be low-risk, and an intermittent ventilation and moisture-proof strategy is adopted. When the risk probability is between the preset low risk threshold and the preset high risk threshold and the real-time health status indicator is within the preset health threshold range, it is determined to be medium risk, and a constant temperature dehumidification and moisture-proof strategy is adopted. When the risk probability is higher than the preset high-risk threshold or the real-time health status indicator is worse than the preset health threshold, it is determined to be high-risk, and a moisture-proof strategy of shutting down dehumidification and isolation protection is adopted. The operational constraints include any number of the following: the rated power range of the equipment, the additional energy consumption limit of the virtual power plant, and the continuous operating time limit of the equipment.

9. The system as described in claim 6, characterized in that, The step of determining the final suitable moisture-proof strategy from the remaining candidate moisture-proof strategies based on the scheduling target issued by the central controller specifically includes: Analyze the scheduling objective to obtain one or more optimization sub-objectives and their priority weights; For each remaining candidate moisture-proof strategy, predict its impact on the one or more optimization sub-objectives after execution, and calculate a comprehensive score based on the priority weights; Compare the overall scores of all remaining candidate moisture-proof strategies, and determine the candidate moisture-proof strategy with the highest overall score as the final suitable moisture-proof strategy.

10. The system as claimed in claim 1, characterized in that, The operating status parameters include at least one of the following: charging and discharging power, operating temperature, and grid connection status; The real-time health status indicators are calculated based on the real-time operating parameters of distributed energy devices monitored by multiple sensors and a predetermined health assessment model. The external environment forecast information is the meteorological data for a specified future period obtained by the virtual power plant central controller from the external meteorological service interface for the deployment site of the distributed energy equipment, including temperature, humidity, and precipitation probability; The moisture prevention strategy library contains multiple moisture prevention strategies, each of which is matched with different risk probability levels, the operating status of distributed energy equipment, and the scheduling objectives of virtual power plants. The moisture-proof operation includes at least one of the following: starting the heater, starting the ventilation device, starting the dehumidifier, and adjusting the operating mode of the distributed energy equipment to generate internal heat; The protected areas of the distributed energy equipment include at least one of the following: the inside of the energy storage battery cabinet, the inside of the photovoltaic inverter cabinet, and the inside of the controller chassis.

Images