Smart energy monitoring analysis system and energy management method

Abstract

The invention discloses a smart energy monitoring and analysis system and an energy management method. The smart energy monitoring and analysis system comprises a multi-mode sensing unit, an edge computing unit, a dynamic transmission unit, an intelligent analysis unit and a closed-loop control unit. All the units interact based on a standardized protocol to form a five-level collaborative closed-loop architecture from data acquisition to strategy execution. According to the system, the data credibility is improved through multi-modal sensing and self-adaptive calibration, the link efficiency is optimized by utilizing a dynamic hierarchical transmission strategy, collaborative optimization calculation is carried out based on a model in which a physical mechanism and data driving are fused, and therefore comprehensive optimization of photovoltaic consumption, energy storage life and energy consumption economy is achieved in a multi-target scene. According to the invention, the overall efficiency and system adaptability of energy management are remarkably improved, and reliable support is provided for stable and efficient operation of the multi-energy complementary system.

Description

Technical Field

[0001] This invention belongs to the field of energy internet technology, specifically a smart energy monitoring and analysis system and energy management method. Background Technology

[0002] With the continuous expansion of global photovoltaic grid connection scale, the rapid popularization of energy storage technology, and the increasing diversification of energy load types, traditional energy monitoring systems have gradually exposed many technical defects in actual operation: On the one hand, traditional systems often use a single sensor for data acquisition, without fully considering the impact of environmental factors such as sudden changes in irradiance, shading, and fluctuations in energy storage temperature on data reliability, leading to large errors in subsequent analysis and calculation, which affects the accuracy of decision-making. On the other hand, the data transmission link uses a fixed link to transmit all types of data. Critical control commands (such as energy storage charging and discharging commands) and routine statistical data (such as historical report data) compete for link resources during transmission, easily causing delays in the transmission of critical commands and affecting system response efficiency. Furthermore, the optimization models of traditional systems are one-sided. Photovoltaic output prediction relies solely on data-driven models, ignoring the physical characteristics constraints of photovoltaic modules (such as the PV characteristic equation), leading to a significant decrease in prediction accuracy when irradiance changes rapidly. Energy storage scheduling optimization focuses only on electricity purchase cost control, failing to incorporate energy storage lifespan degradation into the optimization objective, thus shortening the service life of the energy storage system. Energy consumption optimization lacks scenario-based energy consumption baselines for different operating conditions, resulting in insufficiently targeted energy-saving suggestions that are difficult to effectively tap into energy-saving potential. Simultaneously, traditional systems often employ independent control modes for photovoltaic, energy storage, and energy consumption loads, failing to form a global collaborative control mechanism and thus unable to achieve optimal overall system operation. Therefore, designing a smart energy monitoring and analysis system with multimodal collaborative acquisition, dynamic link transmission, multi-objective optimization calculation, and closed-loop control capabilities has become a key requirement for solving current technical pain points. Summary of the Invention

[0003] The purpose of this invention is to provide a smart energy monitoring and analysis system and an energy management method to solve the problems in data acquisition, transmission, optimization and control of the existing energy monitoring system mentioned in the background art.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A smart energy monitoring and analysis system includes: The multimodal sensing unit is configured to simultaneously collect operating parameters of the photovoltaic power generation system, energy storage system, and energy load; The edge computing unit is connected to the multimodal sensing unit and is configured to perform real-time preprocessing and abnormal state identification on the collected operating parameters. The dynamic transmission unit is configured to dynamically allocate different data transmission link resources to photovoltaic, energy storage and energy consumption scenario data according to their timeliness level. The intelligent analysis unit communicates with the dynamic transmission unit, is deployed in the cloud, and is configured to receive pre-processed and transmitted data, and perform cross-scenario collaborative analysis and optimization calculations based on a model that integrates physical mechanisms and data-driven approaches. The closed-loop control unit is connected to the intelligent analysis unit and is configured to generate specific equipment control commands based on the optimization strategy output by the intelligent analysis unit and send them to the photovoltaic inverter, energy storage converter and energy terminal for execution, while collecting feedback data after the command is executed. Among them, the multimodal sensing unit, edge computing unit, dynamic transmission unit, intelligent analysis unit and closed-loop control unit interact with each other and transmit commands through standardized protocols, together forming a closed-loop control system that achieves the goals of maximizing photovoltaic absorption, optimizing energy storage life and minimizing energy costs.

[0005] According to the above technical solution, the multimodal sensing unit includes a photovoltaic parameter adaptive calibration module, which is configured as follows: A photovoltaic environmental disturbance feature space consisting of irradiance abrupt change amplitude, component temperature gradient, and shading coefficient is constructed, and a nonlinear mapping relationship between it and photovoltaic parameter deviation is established. Based on this deviation, a photovoltaic data credibility assessment mechanism is established, and an exponential function is used to correlate the deviation with credibility. A Bayesian inference framework is adopted to dynamically update the fusion weights of each photovoltaic sensor with the goal of minimizing the photovoltaic power prediction error. The objective is: in Let i be the weight of the i-th sensor. For the power data of the i-th sensor, This is the power estimate after fusion.

[0006] According to the above technical solution, the multimodal sensing unit includes an energy storage parameter degradation prediction module, which is configured as follows: Construct a driving function for energy storage performance degradation, and calculate the degree of degradation by comprehensively considering the effects of SOC fluctuations, cycle count, and temperature. The function is as follows: in, For SOC fluctuation range, To accumulate the number of loops, The value is the average battery temperature; the rest are rated or reference parameters. Will Normalization to Obtain degradation index Set threshold ; According to the above technical solution, the dynamic transmission unit is configured as follows: Define data timeliness types and priorities: photovoltaic ultra-short-term output forecasts and energy storage charging and discharging instructions are classified as emergency; energy consumption anomaly alarms are classified as time-sensitive; historical statistical data are classified as routine. Establish link mapping rules: prioritize 5G / fiber links for emergency cases, allocate LoRa / wireless private network links for time-sensitive cases, and use edge local storage and periodic uploads for regular cases; When the photovoltaic output drops sharply by more than a preset percentage of the rated power, the transmission priority of the associated energy storage charging and discharging commands is automatically increased to the emergency category.

[0007] According to the above technical solution, the intelligent analysis unit includes a photovoltaic output prediction and optimization module, which is configured as follows: A predictive model is constructed using a physical information neural network. The input layer receives data on irradiance, module temperature, wind speed, and historical power output. The physical constraint layer embeds the PV characteristic equation of the photovoltaic module. At this time, it triggers an increase in the SOC sampling frequency and a switch between charging and discharging modes.

[0008] in, For output power, Open circuit voltage, This is the short-circuit current. The fill factor is used; the network training adopts a multi-objective loss function, which combines the mean square error of the predicted and measured forces and the physical constraint residual loss, and balances the two types of loss through weight coefficients; When the rate of change of irradiance exceeds a preset threshold, incremental learning of the model is triggered to ensure that the ultra-short-term prediction error is ≤3% and the short-term prediction error is ≤5%.

[0009] According to the above technical solution, the intelligent analysis unit includes an energy storage scheduling optimization module, which is configured as follows: Construct a multi-objective scheduling model with the goal of minimizing total operating cost. The objective function is: The total operating cost includes the cost of energy storage lifespan loss, electricity purchase cost, and solar curtailment loss; The constraints include: upper and lower limits of SOC operation, energy storage charging and discharging power ≤ rated power, and energy storage charging power meeting photovoltaic consumption requirements; The NSGA-Ⅲ algorithm is used to solve the problem, generating a Pareto optimal scheduling strategy set, and the weights of sub-objectives are dynamically adjusted according to the real-time scenario.

[0010] According to the above technical solution, the intelligent analysis unit includes an energy-saving optimization module, which is configured as follows: Establish corresponding baseline energy consumption models for production, shutdown, and maintenance conditions; Calculate the instantaneous deviation between the actual energy consumption and the current operating condition baseline, and quantify the total energy saving potential by integrating the deviation over time; Generate energy-saving suggestions based on energy consumption scenario types, such as equipment load rate optimization, production staggering, and intelligent control of air conditioning / lighting; Establish an energy-saving effect evaluation mechanism, which evaluates the effect by the ratio of actual energy saving to predicted energy saving. When the evaluation value is less than 80%, iterative optimization of energy-saving suggestions is triggered.

[0011] According to the above technical solution, the closed-loop control unit includes a cross-scenario collaborative control module, which is configured as follows: The control instructions are generated with the goal of achieving global system optimization. The global optimization objective function is: in, For photovoltaic power absorption rate, As an indicator of energy storage degradation, These are the weighting coefficients for degradation indicators and cost items, respectively. The control instructions include: issuing a power limit instruction when photovoltaic output is excessive and energy storage SOC exceeds the threshold; issuing a charging instruction when photovoltaic output surges and load is stable; and issuing a load transfer / reduction instruction when photovoltaic output is insufficient and energy storage SOC is low. The system collects data on photovoltaic power output tracking error, energy storage SOC control error, and load transfer rate as feedback, and adjusts the control parameters when any of these errors exceeds 5%.

[0012] According to the above technical solution, the edge computing unit is configured to: filter photovoltaic parameters in real time to smooth power fluctuations, monitor energy storage parameters in real time and trigger overload protection when SOC / SOH is abnormal, and detect energy consumption parameters in real time to identify load changes and inefficient energy-consuming equipment.

[0013] A smart energy management method includes the following steps: S1, Scenario-based multimodal perception: Collects photovoltaic, energy storage, and energy consumption parameters, adds high-precision UTC timestamps, and unifies the data time step through linear interpolation; S2, Scenario-based data calibration: Calls the photovoltaic parameter adaptive calibration module, the energy storage parameter degradation prediction module, and the energy consumption parameter baseline calibration module to output highly reliable data; S3, Scenario-based intelligent transmission: The dynamic transmission unit selects the path according to the scenario hierarchy, and prioritizes the transmission of photovoltaic prediction data and energy storage charging and discharging instructions; S4, Scenario-based intelligent analysis: The cloud-based photovoltaic output prediction and optimization module generates output predictions, the energy storage scheduling optimization module generates scheduling strategies, and the energy consumption and energy saving optimization module generates energy-saving suggestions. S5, cross-scenario closed-loop control: The closed-loop control unit generates photovoltaic / energy storage / energy consumption control commands through the collaborative control module and sends them to the terminal equipment; it collects execution feedback data and sends it back to each module to optimize parameters, forming a scenario-based closed loop of sensing, calibration, transmission, analysis, control, and feedback.

[0014] Compared with the prior art, the present invention has the following beneficial effects: In this invention, a systematic technological improvement is achieved in the field of energy management through a five-level collaborative architecture and a multi-objective optimization mechanism: In terms of data reliability, multimodal perception and adaptive calibration significantly reduce photovoltaic power fusion errors and enable earlier identification of energy storage degradation risks, effectively extending the overall service life of the energy storage system; in terms of transmission efficiency, a dynamic hierarchical transmission strategy significantly reduces the transmission delay of key commands and significantly improves the utilization rate of link resources; in terms of optimization effect, the system achieves significant progress compared with traditional methods in photovoltaic absorption rate, energy storage operation economy, and energy-saving potential tapping of energy load; in terms of closed-loop reliability, high-precision execution of control commands and rapid iterative updates of models are achieved, significantly enhancing the adaptability to dynamic scenarios, thereby providing strong support for the stable and efficient operation of multi-energy complementary systems. Attached Figure Description

[0015] Figure 1 This is a flowchart of the energy monitoring and analysis system of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 like Figure 1 As shown, this invention achieves end-to-end collaborative management of photovoltaics, energy storage, and energy load by constructing a five-level collaborative architecture encompassing multimodal sensing, edge preprocessing, dynamic transmission, cloud analysis, and closed-loop control. The technical details of each architecture level and its corresponding functional modules are as follows: The system comprises five core functional units, which interact and transmit commands with each other through industry-standard protocols such as ModbusTCP and MQTT, ensuring the stability and compatibility of information transmission.

[0018] The core function of the multimodal sensing unit is to simultaneously collect key operating parameters from photovoltaic, energy storage, and energy consumption scenarios, and to improve the quality of raw data through built-in functional modules. This unit consists of a photovoltaic parameter acquisition module, an energy storage parameter acquisition module, and an energy consumption parameter acquisition module. The photovoltaic parameter acquisition module is equipped with an irradiance sensor with an accuracy of ±5%, a module temperature sensor with an accuracy of ±0.5℃, and a power sensor with an accuracy of ±0.2%. The energy storage parameter acquisition module includes a SOC sensor with an accuracy of ±2%, a battery temperature sensor with an accuracy of ±0.3℃, and a charge / discharge current sensor with an accuracy of ±1%. The energy consumption parameter acquisition module integrates a 0.5-level smart meter and a load classification sensor. To improve data reliability, the unit incorporates a photovoltaic parameter adaptive calibration module and an energy storage parameter degradation prediction module. The former establishes a nonlinear mapping of parameter deviation by constructing a feature space of irradiance abrupt changes, temperature gradients, and shading, and dynamically adjusts the sensor fusion weights using Bayesian inference. The objective function is: in Let i be the weight of the i-th sensor. For the power data of the i-th sensor, This is the power estimate after fusion.

[0019] The latter calculates the degree of energy storage degradation through a degradation driving function and normalizes it to... Degradation index obtained from the interval and set a threshold (e.g., 0.7), when When this occurs, the SOC sampling frequency is automatically increased (from 1 time / minute to 1 time / 10 seconds) or the charge / discharge mode is switched (from full charge / discharge mode to shallow charge / discharge mode); the degradation drive function is as follows: in, For SOC fluctuation range, To accumulate the number of loops, The value is the average battery temperature; the rest are rated or reference parameters.

[0020] The edge computing unit communicates directly with the multimodal sensing unit. Its main function is to receive raw data transmitted by the sensing unit, perform real-time preprocessing and anomaly identification, reduce cloud computing pressure, and improve system response speed. In the data preprocessing stage, the unit uses a Kalman filter algorithm to smooth power fluctuations in photovoltaic power data, a moving average method to eliminate measurement noise in energy storage SOC data, and a threshold filtering method to remove outliers in energy load data. In the anomaly identification stage, the unit presets anomaly judgment thresholds. When it detects a sudden drop in photovoltaic power exceeding 20% ​​of the rated power, an energy storage SOC fluctuation rate exceeding 5% / minute, or a sudden change in energy load exceeding 30%, it immediately triggers a local early warning mechanism and marks the abnormal data as high-priority data for priority transmission to subsequent units.

[0021] The function of the dynamic transmission unit is to formulate a hierarchical transmission strategy based on the differences in data timeliness, optimize the allocation of link resources, and ensure the timeliness of critical data transmission. This unit first categorizes system data into three levels based on timeliness requirements: urgent, timely, and routine. Photovoltaic ultra-short-term output forecast data and energy storage charging / discharging command data are defined as urgent, with a priority factor λ=1.0, and are prioritized for allocation via 5G links (transmission latency <10ms) or fiber optic links. Energy consumption anomaly alarm data and energy storage SOC real-time data are defined as timely, with a priority factor λ=0.9, and are allocated via LoRa links (transmission latency <100ms) or wireless private network links of equivalent bandwidth. Historical output report data and monthly energy consumption statistics are defined as routine, with a priority factor λ<0.5, and are transmitted using edge local storage (e.g., configured with a 128GB industrial-grade SD card) and daily batch uploads to the cloud. Simultaneously, this unit has dynamic adjustment capabilities; when a sudden drop in photovoltaic output exceeds 20% of the rated power, it automatically upgrades the associated energy storage charging / discharging command data from timely to urgent, ensuring rapid transmission of critical control commands.

[0022] The intelligent analysis unit is deployed on a cloud server. Its core function is to construct a multi-objective optimization model based on reliable data preprocessed at the edge, generating strategies for photovoltaic power output prediction, energy storage scheduling optimization, and energy-saving optimization. This unit comprises three core functional modules: the photovoltaic power output prediction and optimization module uses a physical information neural network to construct the prediction model. The network input layer receives data on irradiance (W / m²), module temperature (°C), wind speed (m / s), and historical power output (kW). The physical constraint layer embeds the PV characteristic equation of the photovoltaic module. in For output power, Open circuit voltage, This is the short-circuit current. The fill factor is used. The training process employs a multi-objective loss function to balance prediction data errors and physical constraint residuals. When the rate of change in irradiance exceeds 100 W / (m²・min), the incremental learning process is triggered to ensure that the updated model's ultra-short-term (0-1 hour) prediction error does not exceed 3%, and its short-term (1-4 hour) prediction error does not exceed 5%. The energy storage scheduling optimization module constructs a multi-objective scheduling model with the goal of minimizing total operating costs. The objective function is: in Costs related to energy storage lifespan depletion. For electricity purchase costs, To mitigate photovoltaic (PV) curtailment losses, constraints are set for SOC (State of Charge) operation (20% ≤ SOC ≤ 80%), energy storage charging and discharging power (≤ Rated energy storage power), and PV consumption (charging power ≥ PV output - energy load). The NSGA-Ⅲ multi-objective optimization algorithm is used to solve the model, generating a Pareto optimal scheduling strategy set. The weight coefficients of each sub-objective function are dynamically adjusted based on real-time scenarios. The energy-saving optimization module pre-establishes corresponding baseline energy consumption models for production conditions (e.g., 8:00-20:00), shutdown conditions (e.g., 20:00-8:00), and maintenance conditions (e.g., the 1st of each month). By calculating the instantaneous deviation between actual energy consumption and the current baseline, and combining this with time integration, the overall energy-saving potential is quantified. Specific energy-saving suggestions are generated based on the energy consumption scenario type, such as equipment load rate optimization, production timing staggering, air conditioning temperature setting optimization, or intelligent lighting start / stop. Energy-saving effect evaluation indicators are also established. in To achieve actual energy savings, To predict energy savings, when When the energy consumption is less than 8%, iterative optimization of energy-saving recommendations is triggered.

[0023] The closed-loop control unit communicates with the intelligent analysis unit. Its main function is to transform the comprehensive optimization strategy generated in the cloud into control commands that can be executed by the equipment, and then send them to the photovoltaic inverter, energy storage converter, and energy-consuming terminals for execution. Simultaneously, it collects feedback data after command execution to achieve dynamic optimization of model parameters. This unit generates coordinated control commands with the goal of achieving global system optimization. The global optimal objective function is expressed as: in For photovoltaic power absorption rate, As an indicator of energy storage degradation, These are the weighting coefficients for degradation indicators and cost items, respectively. =0.4、 =0.6 are the weighting coefficients for degradation indicators and cost items, respectively. The specific control logic includes: when photovoltaic output is excessive and the energy storage SOC value is higher than a preset threshold (such as 70%), a power limit command is issued to the photovoltaic inverter. in For photovoltaic power output, This represents the current energy load.

[0024] When photovoltaic output surges and energy load stabilizes, a charging command is sent to the energy storage system: When the photovoltaic output is insufficient and the energy storage SOC value is lower than the preset threshold (such as 30%), a transfer or reduction command is issued to the adjustable energy load. In the feedback optimization stage, the unit collects the photovoltaic output tracking error, energy storage SOC control error and energy load transfer rate as feedback data every minute. When any error exceeds 5%, the generation parameters of the corresponding control command are adjusted in real time to ensure the control effect.

[0025] This embodiment provides a specific implementation of a smart energy management method. Based on the aforementioned smart energy monitoring and analysis system, the corresponding management method achieves full-process energy management through five interconnected steps: First, a multi-modal sensing unit conducts scenario-based data perception and synchronization, collecting core parameters such as irradiance and module temperature for photovoltaic scenarios, SOC and battery temperature for energy storage scenarios, and load power for energy consumption scenarios. Millisecond-level high-precision unified time stamps are added to all collected data to ensure data time consistency. Subsequently, an edge computing unit performs real-time preprocessing and calibration at the edge, filtering and identifying anomalies in the collected data. Simultaneously, it calls the photovoltaic parameter adaptive calibration module, the energy storage parameter degradation prediction module, and the energy consumption parameter baseline calibration module to improve the reliability of the original data through multi-dimensional calibration. Finally, a dynamic transmission unit executes data based on the data's timeliness type and priority. Dynamic hierarchical data transmission allocates optimal communication links to different levels of data, prioritizing the timeliness of transmission of key control commands and real-time early warning data. Then, the cloud-based intelligent analysis unit conducts cloud-based collaborative intelligent analysis, utilizing photovoltaic output prediction optimization modules, energy storage scheduling optimization modules, and energy consumption optimization modules to fuse and analyze cross-scenario data, generating a comprehensive optimization strategy covering photovoltaics, energy storage, and energy consumption. Finally, the closed-loop control unit executes closed-loop collaborative regulation and feedback, transforming the comprehensive optimization strategy into specific control commands for photovoltaic inverters, energy storage converters, and energy consumption terminals, and issuing them for execution. Simultaneously, it collects execution feedback data from terminal devices and transmits it back to the corresponding modules of the intelligent analysis unit and multimodal sensing unit for optimizing model parameters and calibration logic, forming a continuously self-optimizing management closed loop.

[0026] This invention achieves several technological breakthroughs in the field of energy management through a five-level collaborative architecture and a multi-objective optimization mechanism: Regarding data reliability, through multimodal perception and adaptive calibration, the photovoltaic power fusion error is reduced to within 5%, and energy storage degradation prediction can trigger protection mechanisms 1-2 months in advance, effectively extending the lifespan of the energy storage system by 8%-12%; Regarding transmission efficiency, based on a dynamic hierarchical transmission strategy, the transmission delay of key commands is reduced from 500ms in traditional systems to within 10ms, and the utilization rate of link resources is increased by 30%-40%; Regarding optimization effects, the photovoltaic absorption rate is increased by 15%-20% compared to traditional systems, the total operating cost of energy storage is reduced by 10%-15%, and the energy-saving potential of energy loads is tapped at over 85%; Regarding closed-loop reliability, the error in the execution of control commands is controlled within 5%, the model iteration cycle is shortened to 1 hour, and it can quickly adapt to dynamic changes in scenarios, providing strong support for the efficient and stable operation of multi-energy complementary systems.

[0027] Example 2 In this embodiment, to improve the physical reliability and long-term stability of the calibration, the following refined implementation method can be further adopted in the photovoltaic parameter adaptive calibration module: As an optimized and refined implementation of the aforementioned photovoltaic parameter adaptive calibration module, the system can further enhance the interpretability and robustness of the calibration model by introducing a physical reference unit. Specifically, a small representative area (e.g., composed of 2-4 modules) can be selected in the photovoltaic array and used as a reference cell string. This reference string will be subject to the following special management: Environmental protection: Ensure its surface remains clean and unshaded, so that its operating conditions are as close to ideal as possible.

[0028] Independent monitoring: Configure the reference string with independent and more precisely calibrated irradiance and temperature sensors, and monitor their output power. .

[0029] Theoretical calculations: based on the calibration parameters of this reference string under standard test conditions (STC) (such as rated power). Open circuit voltage Short-circuit current Power temperature coefficient (etc.), combined with real-time collected irradiance and temperature The theoretical expected power of a photovoltaic cell under this environment is calculated using formulas based on its physical properties (e.g., a single-diode model or a simplified engineering model). .

[0030] The theoretically predicted power This can serve as a local physical reference for the output power of an ideal photovoltaic unit under current environmental conditions. During calibration, the system can compare this physical reference with the actual measured values ​​of the reference string. Continuous comparisons are performed to verify long-term drift or anomalies in environmental sensors (radiometers, thermometers). Simultaneously, measurements from other common monitoring points in the array can be compared with expected values ​​obtained through spatial interpolation or scaling based on this physical benchmark. In this way, photovoltaic parameter deviations can be more accurately decomposed and understood as: deviations caused by modelable environmental disturbances (abrupt irradiation, temperature gradients), and residual deviations caused by uncontrollable local factors (dust, shading, electrical mismatch). The Bayesian inference framework can integrate these two types of deviation information, dynamically adjusting the fusion weights, so that the calibration process not only relies on statistical data patterns but is also anchored to the underlying physical characteristics of the photovoltaic device.

[0031] Example 3 In the energy storage scheduling optimization module described in Example 1, its objective function includes the energy storage lifetime loss cost. To make the calculation of this cost item more closely reflect the electrochemical mechanism of battery aging and achieve real-time quantification, its implementation method can be further refined: The above-mentioned energy storage lifespan loss cost The calculation can be refined using an aging rate model based on instantaneous stress to more realistically reflect the differentiated impact of different operating conditions on battery life. Specifically: Aging stress factor calibration: A set of instantaneous aging rate coefficients is pre-calibrated by conducting accelerated aging experiments on the type of energy storage battery used or by simulation based on its electrochemical model. This coefficient characterizes the rate of capacity decay or equivalent economic cost per unit of energy throughput of a battery at a specific state of charge (SOC), a specific charge / discharge current (I expressed as a multiple of the rated current, C-rate), and a specific temperature (T). This coefficient can be stored as a multidimensional lookup table or fitted to an empirical function.

[0032] Real-time aging cost calculation: In each discrete time step k within the optimized scheduling cycle, the corresponding instantaneous aging rate coefficient is obtained by looking up a table or by calculation based on the planned charge / discharge current I(k), the predicted SOC trajectory SOC(k), and the battery temperature T(k) determined by the scheduling strategy at the current moment.

[0033] Cost integration: This represents the total lifetime loss cost within the scheduling period. A more accurate estimate can be obtained through the following integral form: in, It is the unit capacity purchase cost of the energy storage system (yuan / kWh). It is the rated voltage. It is approximately the absolute value of energy throughput within time step k. The summation term represents the accumulation of aging damage under different stress states.

[0034] The advantage of this modeling approach lies in transforming energy storage lifetime loss from a macroscopic quantity dependent on post-hoc statistics of a complete charge-discharge cycle into a microscopic process quantity strongly correlated with real-time operating points (SOC, current, temperature). When using algorithms such as NSGA-III for multi-objective optimization, the algorithm can keenly identify and favor scheduling strategies that, while potentially increasing short-term throughput, operate within low-stress ranges (such as suitable SOC ranges and smooth current), thereby optimizing economics while achieving more precise protection of battery life. Essentially, this embeds the electrochemical aging mechanism more deeply into the operations research optimization framework.

[0035] Example 4 The following uses an industrial park (configured with a 2MW photovoltaic array, a 1MWh energy storage system, and a 1.5MW energy load) as an application scenario to illustrate the specific implementation details of the present invention. This implementation case can intuitively demonstrate the actual application effect and technical feasibility of the system.

[0036] In the system deployment phase, the multimodal sensing unit deployment needs to cover all scenarios of photovoltaics, energy storage, and energy consumption. Specifically, 20 sets of integrated irradiance-temperature sensors are evenly deployed in the photovoltaic array area to ensure coverage of all photovoltaic modules; 5 sets of SOC-temperature sensors are deployed inside the energy storage system container to monitor the operating status of the energy storage batteries in real time; and 10 sets of smart meters are deployed on the energy load side of the park (such as production workshops, office buildings, and ancillary facilities) to achieve regional energy consumption data collection. The edge computing unit uses an industrial-grade edge gateway (configured with a quad-core ARM Cortex-A53 CPU and 2GB of memory), deployed in the park's power distribution room, facilitating the reception of data from various sensing modules nearby and reducing transmission losses. The dynamic transmission unit is equipped with a 5G industrial module (supporting standalone (SA) mode) and a LoRa gateway (coverage radius of 1km), with 256GB industrial-grade SSDs for edge local storage to meet the local storage needs of regular data. The cloud-based intelligent analysis unit is deployed on an Alibaba Cloud server (configured with a quad-core CPU and 8GB of memory), using Python programming language combined with the TensorFlow deep learning framework for model training and optimization calculations. The closed-loop control unit is connected to field devices via an RS485 industrial bus, including Huawei SUN2000-2MW photovoltaic inverters, BYD 1MWh energy storage converters, and energy-consuming terminals such as air conditioning controllers and fan controllers in the park, to ensure the effective issuance and execution of commands.

[0037] In terms of typical operating conditions, for the midday sunny day (12:00-13:00, ambient irradiance approximately 1000W / m²), the multimodal sensing unit collected data showing a photovoltaic output of approximately 1.8MW, a storage SOC of approximately 72%, and an energy load of approximately 1.0MW. The edge computing unit filtered the collected data and did not detect any anomalies. Simultaneously, it called the photovoltaic parameter adaptive calibration module to calculate the fusion weights (w1=0.6) and (w2=0.4) of the two sensors, resulting in a fused photovoltaic power of 1.78MW. The dynamic transmission unit then transmitted the photovoltaic ultra-short-term output prediction data (predicting output to remain at 1.7-1000W / m² within one hour). The 0.8MW energy storage charge was marked as an emergency and transmitted to the cloud-based intelligent analysis unit via a 5G link. In the cloud analysis phase, the energy storage scheduling optimization module calculated that the strategy with the lowest total cost was to charge 0.8MW of energy storage (i.e., the difference between photovoltaic output and energy load). The energy-saving optimization module, combined with the production operating condition baseline, proposed to increase the production workshop load rate from 70% to 85%. The closed-loop control unit converted the above strategy into an energy storage charging command (0.8MW) and a load rate adjustment command and issued them for execution. Feedback data collected 10 minutes later showed that the energy storage SOC rose to 76%, the production workshop load rate reached 83%, and the photovoltaic absorption rate reached 100% during this period.

[0038] In response to the sudden change in operating conditions during cloudy weather (15:00, irradiance rapidly decreased from 600W / m² to 200W / m²), the multimodal sensing unit detected that the photovoltaic output dropped from 1.0MW to 0.3MW, the energy storage SOC was approximately 65%, and the energy load was approximately 0.9MW; the edge computing unit identified a sudden drop in photovoltaic power of 40% (exceeding the 20% abnormal threshold), immediately triggered a local warning, and marked the abnormal data as high priority; the dynamic transmission unit automatically upgraded the associated energy storage charging and discharging commands from time-sensitive to emergency, and quickly transmitted them to the cloud via the 5G link; in the cloud intelligent analysis unit, photovoltaic output prediction... The optimization module triggers incremental learning, completing the model parameter update within 10 minutes and predicting that the photovoltaic output will remain at 0.2-0.3MW within 1 hour. The energy storage scheduling optimization module generates a strategy of discharging 0.6MW of energy storage (to meet the difference between energy load and photovoltaic output). The energy-saving optimization module suggests reducing the unnecessary load of the office building by 100kW. The closed-loop control unit issues energy storage discharge instructions (0.6MW) and load reduction instructions. After execution, the feedback data shows that the energy load of the park drops to 0.8MW and the energy storage SOC drops to 62%. During this period, the power supply gap is completely supplemented by the energy storage system, and there is no need to purchase electricity from the grid.

[0039] In the operational effectiveness verification phase, monitoring data from 30 consecutive days of operation of the industrial park system showed that all key performance indicators of the system met the design requirements: the average error of photovoltaic ultra-short-term prediction was 2.5%, with a maximum error of 3.0%; the average error of energy storage SOC control was 2.1%, with a maximum error of 4.8%; the photovoltaic absorption rate reached 92.3%, an increase of 17.2 percentage points compared to the traditional energy monitoring system (75.1%); the energy storage cycle life loss was reduced by 10.5% compared to the fixed charge and discharge strategy; and the energy cost of the park was reduced by 12.8% compared to the previous month before the system deployment, fully verifying the technical feasibility and application value of the invention.

[0040] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0041] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart energy monitoring and analysis system, characterized in that: include: The multimodal sensing unit is configured to simultaneously collect operating parameters of the photovoltaic power generation system, energy storage system, and energy load; The edge computing unit is connected to the multimodal sensing unit and is configured to perform real-time preprocessing and abnormal state identification on the collected operating parameters. The dynamic transmission unit is configured to dynamically allocate different data transmission link resources to photovoltaic, energy storage and energy consumption scenario data based on their timeliness level. The intelligent analysis unit communicates with the dynamic transmission unit, is deployed in the cloud, and is configured to receive pre-processed and transmitted data, and perform cross-scenario collaborative analysis and optimization calculations based on a model that integrates physical mechanisms and data-driven approaches. The closed-loop control unit is connected to the intelligent analysis unit and is configured to generate specific equipment control commands based on the optimization strategy output by the intelligent analysis unit and send them to the photovoltaic inverter, energy storage converter and energy terminal for execution, while collecting feedback data after the command is executed. Among them, the multimodal sensing unit, edge computing unit, dynamic transmission unit, intelligent analysis unit and closed-loop control unit interact with each other and transmit commands through standardized protocols, together forming a closed-loop control system that achieves the goals of maximizing photovoltaic absorption, optimizing energy storage life and minimizing energy costs.

2. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The multimodal sensing unit includes a photovoltaic parameter adaptive calibration module, which is configured as follows: A photovoltaic environmental disturbance feature space consisting of irradiance abrupt change amplitude, component temperature gradient, and shading coefficient is constructed, and a nonlinear mapping relationship between it and photovoltaic parameter deviation is established. Based on this deviation, a photovoltaic data credibility assessment mechanism is established, and an exponential function is used to correlate the deviation with credibility. A Bayesian inference framework is adopted to dynamically update the fusion weights of each photovoltaic sensor with the objective of minimizing the photovoltaic power prediction error. The objective function is: in Let i be the weight of the i-th sensor. For the power data of the i-th sensor, This is the power estimate after fusion.

3. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The multimodal sensing unit includes an energy storage parameter degradation prediction module, which is configured as follows: Construct a driving function for energy storage performance degradation, and calculate the degree of degradation by comprehensively considering the effects of SOC fluctuations, cycle count, and temperature. The function is as follows: in, For SOC fluctuation range, To accumulate the number of loops, The value is the average battery temperature; the rest are rated or reference parameters. Will Normalization to Obtain degradation index Set threshold ;when At this time, it triggers an increase in the SOC sampling frequency and a switch between charging and discharging modes.

4. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The dynamic transmission unit is configured as follows: Define data timeliness types and priorities: photovoltaic ultra-short-term output forecasts and energy storage charging and discharging instructions are classified as emergency; energy consumption anomaly alarms are classified as time-sensitive; historical statistical data are classified as routine. Establish link mapping rules: prioritize 5G / fiber links for emergency cases, allocate LoRa / wireless private network links for time-sensitive cases, and use edge local storage and periodic uploads for regular cases; When the photovoltaic output drops sharply by more than a preset percentage of the rated power, the transmission priority of the associated energy storage charging and discharging commands is automatically increased to the emergency category.

5. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The intelligent analysis unit includes a photovoltaic output prediction and optimization module, which is configured as follows: A predictive model is constructed using a physical information neural network. The input layer receives data on irradiance, module temperature, wind speed, and historical power output. The physical constraint layer embeds the PV characteristic equation of the photovoltaic module. in, For output power, Open circuit voltage, This is the short-circuit current. The fill factor is used; the network training adopts a multi-objective loss function, which combines the mean square error of the predicted and measured forces and the physical constraint residual loss, and balances the two types of loss through weight coefficients; When the rate of change of irradiance exceeds a preset threshold, incremental learning of the model is triggered to ensure that the ultra-short-term prediction error is ≤3% and the short-term prediction error is ≤5%.

6. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The intelligent analysis unit includes an energy storage scheduling optimization module, which is configured as follows: Construct a multi-objective scheduling model with the goal of minimizing total operating cost. The objective function is: The total operating cost includes the cost of energy storage lifespan loss, electricity purchase cost, and solar curtailment loss; The constraints include: upper and lower limits of SOC operation, energy storage charging and discharging power ≤ rated power, and energy storage charging power meeting photovoltaic consumption requirements; The NSGA-Ⅲ algorithm is used to solve the problem, generating a Pareto optimal scheduling strategy set, and the weights of sub-objectives are dynamically adjusted according to the real-time scenario.

7. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The intelligent analysis unit includes an energy-saving optimization module, which is configured as follows: Establish corresponding baseline energy consumption models for production, shutdown, and maintenance conditions; Calculate the instantaneous deviation between the actual energy consumption and the current operating condition baseline, and quantify the total energy saving potential by integrating the deviation over time; Generate energy-saving suggestions based on energy consumption scenario types, such as equipment load rate optimization, production staggering, and intelligent control of air conditioning / lighting; Establish an energy-saving effect evaluation mechanism, which evaluates the effect by the ratio of actual energy saving to predicted energy saving. When the evaluation value is less than 80%, iterative optimization of energy-saving suggestions is triggered.

8. The intelligent energy monitoring and analysis system according to claim 1, characterized in that: The closed-loop control unit includes a cross-scenario collaborative control module, which is configured as follows: The control instructions are generated with the goal of achieving global system optimization. The global optimization objective function is: in, For photovoltaic power absorption rate, As an indicator of energy storage degradation, These are the weighting coefficients for degradation indicators and cost items, respectively. The control instructions include: issuing a power limit instruction when photovoltaic output is excessive and energy storage SOC exceeds the threshold; issuing a charging instruction when photovoltaic output surges and load is stable; and issuing a load transfer / reduction instruction when photovoltaic output is insufficient and energy storage SOC is low. The system collects data on photovoltaic power output tracking error, energy storage SOC control error, and load transfer rate as feedback, and adjusts the control parameters when any of these errors exceeds 5%.

9. A smart energy monitoring and analysis system according to claims 1 to 8, characterized in that: The edge computing unit is configured to: filter photovoltaic parameters in real time to smooth power fluctuations, monitor energy storage parameters in real time and trigger overload protection when SOC / SOH is abnormal, and detect energy consumption parameters in real time to identify load changes and inefficient energy-consuming equipment.

10. A smart energy management method, characterized in that: Using the system according to any one of claims 1-9 includes the following steps: S1, Scenario-based multimodal perception: Collects photovoltaic, energy storage, and energy consumption parameters, adds high-precision UTC timestamps, and unifies the data time step through linear interpolation; S2, Scenario-based data calibration: Calls the photovoltaic parameter adaptive calibration module, the energy storage parameter degradation prediction module, and the energy consumption parameter baseline calibration module to output highly reliable data; S3, Scenario-based intelligent transmission: The dynamic transmission unit selects the path according to the scenario hierarchy, and prioritizes the transmission of photovoltaic prediction data and energy storage charging and discharging instructions; S4, Scenario-based intelligent analysis: The cloud-based photovoltaic output prediction and optimization module generates output predictions, the energy storage scheduling optimization module generates scheduling strategies, and the energy consumption and energy saving optimization module generates energy-saving suggestions. S5, cross-scenario closed-loop control: The closed-loop control unit generates photovoltaic / energy storage / energy consumption control commands through the collaborative control module and sends them to the terminal equipment; it collects execution feedback data and sends it back to each module to optimize parameters, forming a scenario-based closed loop of sensing, calibration, transmission, analysis, control, and feedback.

Images