An intelligent power consumption monitoring system

Abstract

The present invention discloses an intelligent power consumption monitoring system, which relates to the technical field of intelligent energy management. It includes collecting environmental parameters, power load status, and energy consumption data in real time and fusing them with historical energy consumption records, real-time electricity price information, and weather forecast data to construct a multi-dimensional dynamic energy consumption model to predict the energy consumption trend; formulating a dynamic adjustment strategy according to the energy consumption trend predicted by the multi-dimensional dynamic energy consumption model; dynamically monitoring the power grid status, dynamically analyzing the power grid status according to the predicted energy consumption trend and real-time power grid conditions, and determining the optimal load distribution strategy for each power consumption unit; automatically executing the load balancing strategy through intelligent control devices; monitoring the actual effect after implementing the strategy, and evaluating the execution efficiency and energy-saving effect of the strategy. The present invention effectively improves the power grid response speed, energy utilization rate, and energy conservation and emission reduction effect, and promotes the practice of intelligent energy.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent energy management, and in particular to an intelligent power consumption monitoring system. Background Art

[0002] In recent years, with the continuous growth of global energy demand and the increasing awareness of environmental protection, intelligent energy management has become one of the key areas of scientific and technological and social development. Traditional energy management systems mostly focus on passive data recording and basic analysis, lacking the ability to dynamically respond to complex environmental changes and user needs. Therefore, it is particularly important to develop an intelligent power consumption monitoring system that can autonomously learn, adjust in real time, and have high adaptability.

[0003] Under the existing technical framework, most systems fail to fully utilize the dynamic correlation between environmental factors and power load status, resulting in limited accuracy of prediction models. Strategy formulation often relies on fixed rules and simple prediction models, lacking the ability to adaptively adjust to the real-time state of the power grid, and it is difficult to effectively cope with the challenges of large energy consumption fluctuations and rapid demand changes. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent power consumption monitoring system to solve the problems of traditional power consumption monitoring systems in data processing, energy consumption prediction, dynamic strategy adjustment, and adaptive load control.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides an intelligent power consumption monitoring system, which includes:

[0008] Data acquisition and preprocessing module: Real-time collect environmental parameters, power load status, and energy consumption data through environmental sensors and smart meters and perform preprocessing;

[0009] Multi-dimensional dynamic energy consumption model construction module: Integrate the preprocessed data with historical energy consumption records, real-time electricity price information, and weather forecast data to construct a multi-dimensional dynamic energy consumption model to predict energy consumption trends;

[0010] Dynamic adjustment strategy formulation module: Use machine learning algorithms to formulate dynamic adjustment strategies according to the energy consumption trends predicted by the multi-dimensional dynamic energy consumption model;

[0011] Intelligent load balancing module with adaptive learning: Adopt reinforcement learning technology to dynamically monitor the power grid status, dynamically analyze the power grid status according to the predicted energy consumption trends and real-time power grid conditions, and determine the optimal load distribution strategy for each power consumption unit;

[0012] Policy Execution and Control Module: Receives control instructions from the intelligent load balancing module and automatically executes the load balancing policy through intelligent control devices;

[0013] Effect Evaluation and Feedback Optimization Module: Monitors the actual effects after implementing the policy, collects power grid operation status, energy consumption data, and user feedback, and evaluates the execution efficiency and energy-saving effectiveness of the policy.

[0014] As a preferred embodiment of the intelligent power consumption monitoring system described in the present invention, wherein: the environmental parameters include temperature, humidity, and light;

[0015] The power load status includes current, voltage, power, and cumulative energy consumption.

[0016] As a preferred embodiment of the intelligent power consumption monitoring system described in the present invention, wherein: preprocessing refers to the preliminary screening of the collected data, removing outliers that significantly exceed the physical range and equipment specifications, and smoothing the data.

[0017] As a preferred embodiment of the intelligent power consumption monitoring system described in the present invention, wherein: the preprocessed data is fused with historical energy consumption records, real-time electricity price information, and weather forecast data. The specific operation steps are as follows:

[0018] Feature extraction is performed on the preprocessed environmental parameters, power load status, and cumulative energy consumption to form a comprehensive feature vector, which is integrated with historical energy consumption records, real-time electricity price information, and weather forecast data. Based on the integrated data, a fusion feature is constructed, and the expression is:

[0019] F t =w 1 ·MA k (H t-k:t )+w 2 ·E t +w 3 ·P t +w 4 ·W t +w 5 ·ES α (T t-1:t-k );

[0020] Wherein, F t represents the comprehensive feature vector at time point t, w 1 represents the weight factor of historical energy consumption, t represents the current time point, w 2 represents the weight factor of real-time energy consumption, w 3 represents the weight factor of real-time electricity price, w 4 represents the weight factor of external environmental factors, w 5 represents the weight factor of the time series trend, MA k (Ht-k:t ) represents the moving average of historical energy consumption over the most recent k time points, E t represents the real-time energy consumption value, P t represents the real-time electricity price, W t represents the environmental parameter vector, ESα(T t-1:t-k ) represents the exponentially smoothed value of the time series trend, α represents the smoothing coefficient, T t-1:t-k represents the time series data from t - 1 to t - k.

[0021] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, wherein: construct a composite influence function I t , and use the multiple regression analysis method to quantify the influence of each dimension on energy consumption. The expression is:

[0022]

[0023] wherein, I t represents the value of the composite influence function at time point t, β d represents the set of regression coefficients in the multiple regression analysis, K(x,σ 2 ) represents the Gaussian kernel function, x represents the distance in the feature space, σ 2 represents the variance parameter.

[0024] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, wherein: combine the fusion features and the multi-dimensional influence analysis of the composite influence function on energy consumption to construct a dynamic energy consumption prediction model and output the dynamic energy consumption prediction value. The expression is:

[0025]

[0026] wherein, E p (t + h) represents the energy consumption prediction value at time point t + h, h represents the predicted future time span, m represents the order of the piecewise polynomial, c represents the amplitude parameter in the prediction model, T represents the period parameter in the prediction model represents the sine function term, Norm(.) represents the normalization function, a j and b j represent the key parameters in the polynomial regression.

[0027] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, wherein: use the machine learning algorithm to formulate a dynamic adjustment strategy according to the dynamic energy consumption prediction value. The specific operation steps are as follows:

[0028] Analyze the energy consumption prediction curve provided by the dynamic energy consumption prediction value, identify the peak and trough periods of energy consumption, and the key nodes of the energy consumption change rate;

[0029] Based on historical data, evaluate the energy consumption response sensitivity of each electrical device, define the device priority matrix and the minimum power margin for the safe operation of the system;

[0030] Construct an objective function for minimizing the cost of energy consumption and device start-stop costs, and the expression is:

[0031]

[0032] Among them, C t represents the minimized cost, N represents the total number of devices, i represents the device index variable, P c (t) represents the adjusted real-time electricity price, E i (t) represents the actual energy consumption of device i at time t, φ i represents the start-stop cost coefficient of device i, β represents the coefficient related to the device's adjusted energy consumption cost, γ represents the exponential parameter, λ represents the adjustment frequency cost factor, and V(t) represents the frequency of policy adjustment;

[0033] Use a deep network to learn the best policy by interacting with the environment with the goal of minimizing the cost;

[0034] Deploy the best policy to the intelligent power consumption management system, dynamically adjust the operating status and power level of the device according to the prediction results, reduce the energy consumption in advance before the energy consumption peak, and make reasonable use of resources during the trough.

[0035] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, among them: adopt reinforcement learning technology to dynamically monitor the power grid status, dynamically analyze the power grid status according to the predicted energy consumption trend and real-time power grid conditions, and determine the best load distribution strategy for each power consumption unit. The specific operation steps are as follows:

[0036] Define the state vector S t =[E t , E p (t + h), P t , W t , D t , where E t represents the real-time energy consumption value, and W t represents the environmental parameter vector;

[0037] Create an action space that contains all device operation instructions A t ;

[0038] Design a reward function, and the expression is:

[0039]

[0040] Among them, Δf i represents the deviation of the system frequency from the standard value, and ε represents the scaling coefficient;

[0041] Received state vector S t to generate the expected Q-value for each possible action a i with the expression:

[0042]

[0043] where R t+1 represents the immediate feedback obtained at time step t after taking action A t and a′ represents one of all actions in the next state S t+1 to calculate the maximum expected return of the next state;

[0044] Execute the selected action and observe the new state S t+1 after execution and the immediate reward R t and adjust the strategy according to the actual power grid feedback;

[0045] Apply the learned strategy to the intelligent power consumption management system to dynamically adjust the power level and start / stop status of the devices.

[0046] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, wherein: receive control instructions from the intelligent load balancing module and automatically execute the load balancing strategy through the intelligent control device, and the specific operation steps are as follows:

[0047] The monitoring system receives control instructions from the intelligent load balancing module for the specific operation requirements of each device;

[0048] Based on the working state of the current device and the received control instructions, perform the status;

[0049] According to the device status evaluation results, prioritize all devices to be adjusted, consider factors such as energy consumption saving potential, device response speed, and mutual influence between devices, and formulate an execution sequence;

[0050] According to the established execution sequence, send precise control signals to each device through the intelligent controller;

[0051] Collect the real-time data after the device adjustment, compare it with the prediction model, and evaluate the strategy execution effect. If a deviation is found, promptly start the feedback error correction model and adjust based on the deviation between the predicted value and the actual value until the expected goal is achieved, with the expression:

[0052]

[0053] where ΔP c,i represents the correction power adjustment amount of device i, k p represents the proportional control coefficient, k i represents the integral control coefficient, Ep,i (t + h) represents the predicted energy consumption of device i at time t + h, E a,i (t + h) represents the actual energy consumption of device i at time t + h, t 0 represents the starting time point, τ represents the integration variable, representing each time point within the time range from a certain initial time t 0 to the current time t.

[0054] As a preferred solution of the intelligent power consumption monitoring system described in the present invention, wherein: monitor the actual effect after implementing the strategy, collect the grid operation status, energy consumption data and user feedback, and evaluate the execution efficiency and energy-saving effect of the strategy. The specific operation steps are as follows:

[0055] Continuously collect and integrate the real-time operation status data of the power grid;

[0056] Establish a comprehensive evaluation index system including the strategy response time, execution accuracy rate and energy-saving efficiency. The expression is:

[0057]

[0058] where ηe represents the energy-saving efficiency, Ep_t represents the total predicted energy consumption before implementing the strategy, Ea_b represents the total actual energy consumption before implementing the energy-saving strategy, and Ea_a represents the total actual energy consumption after implementing the energy-saving strategy;

[0059] Based on time series analysis, combined with the long-term trend and seasonal changes of energy conservation, construct a quantitative model of energy-saving effect. The expression is:

[0060]

[0061] where ΔEs,v represents the predicted seasonal energy consumption increment at time point v, and B represents the length of the analysis time window;

[0062] Design questionnaires or online surveys to collect users' satisfaction evaluations after implementing the strategy and convert them into quantitative indicators;

[0063] Introduce environmental adaptability indicators to evaluate the adaptability and stability of the strategy under different environmental conditions;

[0064] Combined with the above evaluation results, construct a feedback optimization model to dynamically adjust the strategy parameters and device operation instructions. The expression is:

[0065] J = η e + a·U s + b·A e

[0066] where J represents the overall benefit function to be maximized, Us represents the satisfaction score, A eIndicates the environmental adaptability index, a represents the system target trade-off coefficient, and b represents the user requirement trade-off coefficient;

[0067] According to the results of the feedback optimization model, adjust the policy parameters and redeploy them to the intelligent power consumption management system.

[0068] The beneficial effects of the present invention are as follows: real-time collection and preprocessing of data such as environmental parameters and power loads to ensure data accuracy; construction of a multi-dimensional energy consumption model by integrating historical and real-time information to achieve accurate energy consumption prediction; formulation of a dynamic adjustment strategy using machine learning to optimize energy distribution; adoption of reinforcement learning to dynamically balance the load to ensure the stability and efficiency of the power grid; the system automatically executes the strategy and monitors the effect, and continuously optimizes through user feedback to form a closed-loop management, effectively improving the power grid response speed, energy utilization rate, and energy conservation and emission reduction effects, and promoting the practice of smart energy. Description of the Drawings

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0070] Figure 1 It is the flowchart of the intelligent power consumption monitoring system in Embodiment 1.

[0071] Figure 2 It is the diagram of implementing the optimal load distribution strategy in Embodiment 1. Detailed Embodiments

[0072] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification.

[0073] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar promotions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0074] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that mutually excludes other embodiments.

[0075] Embodiment 1, refer to Figure 1 and Figure 2, which is the first embodiment of the present invention. This embodiment provides an intelligent power consumption monitoring system, including the following steps:

[0076] S1. Real-time collect environmental parameters, power load status and energy consumption data through environmental sensors and smart meters and perform preprocessing.

[0077] Furthermore, the environmental parameters include temperature, humidity and light;

[0078] The power load status includes current, voltage, power and cumulative energy consumption.

[0079] Furthermore, the preprocessing refers to performing preliminary screening on the collected data, removing outliers that significantly exceed the physical range and equipment specifications, and smoothing the data.

[0080] Furthermore, perform range checks on power load data such as current, voltage, and power to ensure that the data falls within the normal operating range of the equipment. For example, set the reasonable range of current as I min to I max , if the collected current value I t is not within this range, it is marked as an outlier;

[0081] Apply the difference method or the Z-score method to identify mutation points in the data. For example, calculate the difference ∣X t -X t-1 ∣, if the difference exceeds the set threshold δ or the Z-score is greater than the predetermined standard (usually 3), it is regarded as an abnormal sudden increase and is removed or corrected;

[0082] Ensure that the timestamps of all data sources are consistent, align the data misalignment caused by the acquisition time difference, and handle missing or misaligned data points through interpolation (such as linear interpolation) or the nearest neighbor filling method to ensure the time consistency of the data sequence;

[0083] Determine the value of the smoothing factor α, generally selected according to the data characteristics and requirements. For scenarios that require a faster response to recent changes, α can be appropriately increased (close to 1), while for scenarios that need to consider more historical trends, a smaller value (close to 0 but not equal to 0) is selected;

[0084] When applying EMA for the first time, it is necessary to set the initial value of S 0 , take the initial data point X 0 as S 0 , or use the mean value of the data set as the starting smoothing value to reduce the deviation in the initial stage;

[0085] According to the formula S t =α·X t +(1-α)·St-1 Smooth the data at each time point in sequence;

[0086] During the application of EMA, pay attention to monitoring the smoothing effect to ensure that high-frequency noise is effectively suppressed while retaining the overall trend and periodic characteristics of the data;

[0087] According to the performance of the preliminarily smoothed data, adopt advanced smoothing techniques such as double moving average (such as the combination of simple moving average and EMA) to better balance noise suppression and signal retention until a satisfactory smoothing effect is achieved.

[0088] It should be noted that by real-time collecting environmental parameters, power load status and energy consumption data through environmental sensors and smart meters, the real-time monitoring of the power grid operation environment and power consumption situation is realized, providing basic data support for subsequent energy consumption management and optimization strategies. The collected data is preliminarily screened to remove outliers that significantly exceed the physical range and equipment specifications. By setting reasonable data ranges, the validity of the data is guaranteed, avoiding wrong decisions caused by sensor false alarms or data transmission errors. The difference method or Z-score method is used to identify and process mutation points in the data. By setting thresholds or Z-score criteria, the noise and abnormal mutations in the data are effectively filtered, maintaining the continuity and stability of the data sequence.

[0089] S2. Integrate the preprocessed data with historical energy consumption records, real-time electricity price information and weather forecast data. The specific operation steps are as follows:

[0090] Extract features from the preprocessed environmental parameters, power load status and cumulative energy consumption to form a comprehensive feature vector F, and integrate it with historical energy consumption records, real-time electricity price information and weather forecast data. Based on the integrated data, construct a fusion feature, and the expression is:

[0091] F t = w 1 ·MA k (H t-k:t ) + w 2 ·E t + w 3 ·P t + w 4 ·W t + w 5 ·ES α (T t-1:t-k );

[0092] Where, F t represents the comprehensive feature vector at time point t, w 1 represents the weight factor of historical energy consumption, t represents the current time point, w 2 represents the weight factor of real-time energy consumption, w3 The weight factor representing the real-time electricity price, w 4 The weight factor representing the external environmental factors, w 5 The weight factor representing the time series trend, MA k (H t-k:t ) represents the moving average of historical energy consumption over the most recent k time points, used to capture the short-term trend of energy consumption, E t Represents the real-time energy consumption value, the instantaneous energy consumption reading at time point t, P t Represents the real-time electricity price, reflecting the variation of electricity cost over time, W t Represents the environmental parameter vector, the environmental parameter values at time point t, such as temperature, humidity, etc., ESα(T t-1:t-k ) represents the exponentially smoothed value of the time series trend, α represents the smoothing coefficient, determining the influence degree of new data on the model, T t-1:t-k Represents the time series data from t - 1 to t - k.

[0093] Furthermore, construct a composite influence function I t , and use the multiple regression analysis method to quantify the influence of each dimension on energy consumption. The expression is:

[0094]

[0095] Among them, I t represents the value of the composite influence function at time point t, integrating the quantified influence of each dimension on energy consumption, β d represents the set of regression coefficients in the multiple regression analysis, used to quantify the predictive influence of historical data on future energy consumption, K(x,σ 2 ) represents the Gaussian kernel function, x represents the distance in the feature space, σ 2 represents the variance parameter, controlling the width of the Gaussian distribution and affecting the degree of attention of the influence function to nearby data points.

[0096] Furthermore, combining the fused features and the multi-dimensional influence analysis of the composite influence function on energy consumption, construct a dynamic energy consumption prediction model, and output the dynamic energy consumption prediction value. The expression is:

[0097]

[0098] Among them, E p (t + h) represents the energy consumption prediction value at time point t + h, h represents the predicted future time span, m represents the order of the piecewise polynomial, determining the fitting ability of the model to non-linear relationships, c represents the amplitude parameter in the prediction model, and T represents the period parameter in the prediction model Represents the sine function term, introducing periodic fluctuations to reflect regular changes such as seasonality. Norm(.) represents the normalization function, ensuring that the predicted values are within a reasonable numerical range, avoiding extreme predicted values, and improving the practicality and stability of the model, a j and b j Represent the key parameters in polynomial regression, jointly determining how to incorporate the weighted sum of historical impacts and time factors into the future prediction of energy consumption, enabling the model to more accurately capture the trend of energy consumption changes over time and across various influencing factors.

[0099] It should be noted that from data integration, feature extraction to multi-dimensional impact analysis and prediction model construction, the energy efficiency management and prediction capabilities of the intelligent power consumption system have been systematically improved. It can not only accurately predict future energy consumption trends but also carefully analyze multi-dimensional factors affecting energy consumption, providing strong decision-making support for formulating energy-saving strategies and optimizing power grid dispatching.

[0100] S3. Use machine learning algorithms to formulate dynamic adjustment strategies based on dynamic energy consumption prediction values. The specific operation steps are as follows:

[0101] Analyze the energy consumption prediction curve provided by the dynamic energy consumption prediction value, identify peak and trough periods of energy consumption, and key nodes of the energy consumption change rate;

[0102] Based on historical data, evaluate the energy consumption response sensitivity of each electrical device, and define the device priority matrix and the minimum power margin for the safe operation of the system;

[0103] Construct an objective function for minimizing the cost of energy consumption and equipment start-stop costs. The expression is:

[0104]

[0105] Among them, C t Represents the minimized cost, comprehensively considering energy consumption, equipment start-stop costs, and the frequency cost of strategy adjustment. N represents the total number of devices, i represents the device index variable, P c (t) represents the adjusted real-time electricity price, reflecting the electricity price actually paid at time t, which is adjusted according to time periods (such as peak, off-peak) to reflect the changes in electricity costs during different time periods, E i (t) represents the real-time energy consumption of device i at time t. The higher the energy consumption, the higher the cost paid at the given electricity price, φ iLet $\alpha_i$ denote the start-stop cost coefficient of device $i$, which reflects the fixed cost or marginal cost of starting or stopping the device once. $\beta$ represents the coefficient related to the energy consumption cost of device adjustment, affecting the cost of start-stop or large-scale energy consumption adjustment. $\gamma$ represents an exponential parameter used to control the growth rate of the energy consumption adjustment cost with the adjustment amplitude. When $\gamma > 1$, it indicates that the adjustment cost increases rapidly with the increase of the adjustment amplitude, emphasizing the importance of avoiding large-scale adjustments. $\lambda$ represents the adjustment frequency cost factor, used to balance the cost brought by frequent strategy adjustments. $V(t)$ represents the frequency of strategy adjustment, indicating the number or frequency of adjusting the operating state (such as start-stop or power setting) of the device at time $t$.

[0106] Use a deep network to minimize the cost and learn the optimal strategy by interacting with the environment.

[0107] Deploy the optimal strategy to the intelligent power consumption management system, dynamically adjust the operating state and power level of the device according to the prediction results, reduce the energy consumption in advance before the energy consumption peak, and reasonably utilize resources during the low valley.

[0108] It should be noted that by analyzing the energy consumption prediction curve provided by the multi-dimensional dynamic energy consumption model, an in-depth understanding of the energy consumption pattern is achieved, especially identifying the energy consumption peak and low valley periods, which provides a time window for formulating an efficient energy management strategy. Based on historical data, the energy consumption response sensitivity of each electrical device is evaluated, and the device priority matrix and the minimum power margin for the safe operation of the system are defined. This process ensures the scientificity and security of the device management strategy. Construct a cost objective function that minimizes the energy consumption and the start-stop cost of the device, comprehensively considering the energy consumption, the start-stop cost of the device, and the frequency cost of strategy adjustment. Use a deep network to minimize the cost as the goal, learn the optimal strategy by interacting with the environment, and deploy the optimal strategy to the intelligent power consumption management system. This process realizes the intelligence and automation of strategy optimization.

[0109] S4. Adopt reinforcement learning technology to dynamically monitor the power grid state, dynamically analyze the power grid state according to the predicted energy consumption trend and real-time power grid conditions, and determine the optimal load distribution strategy for each power consumption unit. The specific operation steps are as follows:

[0110] Define the state vector $S$ t $ = [E$ t , E$ p $(t + h), P$ t , W$ t , D$ t $, where $E$ t represents the real-time energy consumption value, $W$ t represents the environmental parameter vector, including temperature, humidity, and light, and $D$ t represents the device working state vector, indicating whether each device is in the on state.

[0111] The created action space contains all device operation instructions A t , such as changing the device power level or switch state;

[0112] Design the reward function, with the expression:

[0113]

[0114] where, Δf i represents the deviation of the system frequency from the standard value. After summation and taking the inverse, the larger value indicates greater stability. ε represents the scaling coefficient, meaning that the higher the cost, the larger the value deducted from the reward;

[0115] Receive the state vector S t , with dimensions matching the state space dimension, and generate the expected Q-value for each possible action a i , with dimensions the same as the action space, and the expression is:

[0116]

[0117] where, R t+1 represents the immediate feedback obtained at time step t after taking action A t . It reflects the goodness or badness after executing a certain action and directly affects the optimization direction of the strategy. a′ represents one of all actions in the next state S t+1 , and calculates the maximum expected return of the next state;

[0118] Execute the selected action and observe the new state S t+1 and the immediate reward R t , and adjust the strategy according to the actual power grid feedback;

[0119] Apply the learned strategy to the intelligent power consumption management system to dynamically adjust the power level and start / stop state of the device.

[0120] It should be noted that by defining the state vector, a comprehensive characterization of the current state and future prediction of the power grid is achieved. By creating an action space containing all device operation instructions, fine control of the operation of power consumption units within the power grid is realized. The designed reward function achieves a comprehensive consideration of the optimization goal by weighing the cost, system frequency stability, and voltage fluctuation. By receiving the state vector to generate the expected value of each possible action and adjusting the strategy according to the immediate feedback and maximum expected return, continuous optimization and self-learning of the strategy are realized.

[0121] S5. Receive the control instruction from the intelligent load balancing module and automatically execute the load balancing strategy through the intelligent control device. The specific operation steps are as follows:

[0122] The monitoring system receives control instructions from the intelligent load balancing module for the specific operation requirements of each device, including power adjustment values and switch status change instructions;

[0123] Based on the current working status of the device and the received control instructions, perform status evaluation, which includes but is not limited to the immediate energy consumption of the device, the difference between the current power level and the instruction target, and the estimated device response time;

[0124] According to the device status evaluation results, prioritize all devices to be adjusted, consider the potential for energy consumption savings, device response speed, and mutual influence factors between devices, and formulate an execution sequence;

[0125] According to the established execution sequence, send precise control signals to each device through the intelligent controller, including power adjustment instructions or switch instructions;

[0126] Collect the real-time data after device adjustment, including energy consumption changes and device status feedback, compare it with the prediction model, and evaluate the execution effect of the strategy. If a deviation is found, promptly start the feedback error correction model and make adjustments based on the deviation between the predicted value and the actual value until the expected target is reached. The expression is:

[0127]

[0128] Among them, ΔP c,i represents the corrected power adjustment amount of device i, k p represents the proportional control coefficient, k i represents the integral control coefficient, E p,i (t + h) represents the predicted energy consumption of device i at time t + h, E a,i (t + h) represents the actual energy consumption of device i at time t + h, t 0 represents the starting time point, τ represents the integration variable, representing each time point within the time range from a certain initial time t 0 to the current time t.

[0129] It should be noted that by receiving the control instructions of the intelligent load balancing module through the monitoring system, the effective communication between the intelligent power consumption system and the intelligent control center is realized, ensuring real-time and accurate instruction transmission. Based on the device working status and control instructions, status evaluation is performed, realizing the comprehensive consideration of device adjustment requirements. According to the evaluation results, an execution sequence is formulated and precise control signals are sent through the intelligent controller, ensuring the orderly execution of device adjustment, improving the coordination and efficiency of the entire system. Collecting the real-time data after device adjustment and comparing it with the prediction model, and enabling the feedback error correction model for adjustment, realizing the real-time monitoring and dynamic adjustment of the execution effect of the strategy.

[0130] S6. Monitor the actual effects after implementing the strategy, collect power grid operation status, energy consumption data, and user feedback, and evaluate the execution efficiency and energy-saving effectiveness of the strategy. The specific operation steps are as follows:

[0131] Continuously collect and integrate real-time power grid operation status data, including but not limited to current, voltage, power, and cumulative energy consumption. At the same time, collect environmental parameters (temperature, humidity, light) and real-time electricity price information and user feedback, especially information about electricity usage experience and energy-saving perception;

[0132] Establish a comprehensive evaluation index system including strategy response time, execution accuracy rate, and energy-saving efficiency. The expression is:

[0133]

[0134] Among them, ηe represents the energy-saving efficiency, Ep_t represents the total predicted energy consumption before implementing the strategy, Ea_b represents the total actual energy consumption before implementing the energy-saving strategy, and Ea_a represents the total actual energy consumption after implementing the energy-saving strategy;

[0135] Based on time series analysis, combined with the long-term trend and seasonal changes of energy conservation, construct a quantitative model of energy-saving effect. The expression is:

[0136]

[0137] Among them, ΔEs,v represents the predicted seasonal energy consumption increment at time point v, B represents the length of the analysis time window, in months or quarters, to ensure full consideration of periodic changes;

[0138] Design questionnaires or online surveys to collect user satisfaction evaluations after the implementation of the strategy, covering aspects such as convenience, comfort, and economic burden, and convert them into quantitative indicators;

[0139] Introduce environmental adaptability indicators to evaluate the adaptability and stability of the strategy under different environmental conditions, considering the impact of environmental parameters on the accuracy of energy consumption prediction and the energy-saving effect;

[0140] Combined with the above evaluation results, construct a feedback optimization model to dynamically adjust strategy parameters and equipment operation instructions. The expression is:

[0141] J=η e +a·U s +b·A e

[0142] Among them, J represents the overall benefit function to be maximized, Us represents the satisfaction score, A e represents the environmental adaptability indicator, a represents the system target trade-off coefficient, and b represents the user demand trade-off coefficient;

[0143] According to the results of the feedback optimization model, the strategy parameters are adjusted and redeployed to the smart power management system.

[0144] It should be noted that through comprehensive data collection, the comprehensiveness and accuracy of strategy evaluation are ensured, which can timely reflect the actual operating status of the power grid after the implementation of the strategy and the real experience of users, establish a comprehensive evaluation index system including energy-saving efficiency, optimize the energy-saving effect of the strategy, and clearly demonstrate the efficiency of the energy-saving strategy in practical application, which helps to quickly identify the effectiveness of the energy-saving strategy and provide a quantitative reference standard for continuous improvement. Time series analysis is used in combination with seasonal changes to construct a quantitative model of energy-saving effects, which makes the energy-saving effect evaluation more accurate and comprehensive, helps to discover and utilize seasonal patterns, further optimize energy utilization strategies, and improve the sustainability and efficiency of energy saving.

[0145] In summary, the present invention ensures data accuracy by real-time collection and preprocessing of environmental parameters, power load and other data; integrates historical and real-time information to build a multi-dimensional energy consumption model to achieve accurate energy consumption prediction; uses machine learning to formulate dynamic adjustment strategies to optimize energy distribution; uses reinforcement learning to dynamically balance loads to ensure grid stability and efficiency; the system automatically executes strategies and monitors effects, continuously optimizes through user feedback, forms a closed-loop management, effectively improves grid response speed, energy utilization, and energy conservation and emission reduction effects, and promotes smart energy practices.

[0146] Example 2

[0147] Referring to Table 1, which is the second embodiment of the present invention, experimental simulation data of the smart electricity monitoring system is provided to further verify the advancement of the present invention.

[0148] First, a set of virtual experimental environments were configured to simulate the typical electricity consumption patterns of different industries and residential areas. Smart meters and environmental sensors were installed in the experimental environments to collect data in real time with high precision.

[0149] During the implementation process, the data preprocessing stage adopted advanced outlier detection algorithms, such as the IQR (interquartile range) rule, to eliminate measurements beyond the normal range, and applied a moving average filter for smoothing to ensure data quality. Subsequently, the data was integrated into a multi-dimensional feature vector F, which included the combined effects of historical energy consumption, real-time energy consumption, electricity prices, weather conditions, and time series trends, and a comprehensive energy consumption prediction model foundation was constructed using the weighted fusion method.

[0150] Furthermore, we applied multiple regression analysis to construct a composite impact function It, which quantified the comprehensive impact of each dimension on energy consumption through a Gaussian kernel function, improving the prediction accuracy of the model. Based on this, a dynamic energy consumption prediction model was developed. In the model, not only piecewise polynomials and sine function terms were incorporated, but also the reasonable range of predicted values was ensured through normalization processing, significantly enhancing the flexibility and practicality of the model.

[0151] In the strategy formulation stage, through deep network learning, the system can dynamically adjust the device operation strategy according to the predicted energy consumption trend curve and optimize the cost function. This process fully considers the real-time electricity price, device response cost, and adjustment frequency, achieving a balance between cost minimization and system stability. At the same time, reinforcement learning technology is introduced, enabling the system to dynamically adjust the load distribution strategy according to the grid state, further improving the energy use efficiency.

[0152] Finally, through intelligent control devices to execute the strategy and real-time monitoring and feedback, closed-loop control is achieved. When there is a difference between the predicted value and the actual energy consumption, the system immediately activates the feedback error correction model and adjusts the device power according to the proportional-integral control principle to ensure the achievement of the energy-saving goal, as shown in the following table:

[0153] Table 1 Experimental Record Table of Intelligent Power Consumption Monitoring System

[0154]

[0155] By comparing the "predicted energy consumption" and "actual energy consumption" in the experimental data and combining with the "energy-saving efficiency" column, it can be clearly seen that the intelligent power consumption monitoring system of the present invention not only accurately predicts the energy consumption trend, but also the average energy-saving efficiency reaches 0.22% in practical applications, indicating the significant effect of the system in reducing energy waste. Especially when the environmental parameters fluctuate greatly, the system can respond quickly, effectively avoiding overload and excessive energy consumption, achieving at least 5% additional energy-saving gain compared with the traditional static management method.

[0156] By analyzing the difference between the "cumulative energy consumption" and "actual energy consumption", the energy-saving effect of the system in different time periods and its adaptability to environmental factors are also verified. Especially during the low-load period at night, through intelligent scheduling, the system effectively utilizes low-cost electricity, and the overall energy-saving efficiency is about 10% higher than that during the day, demonstrating the excellent adaptability and energy-saving potential of the system under complex working conditions.

[0157] In summary, through a series of innovative data processing, model construction, and strategy optimization methods, the present invention significantly improves the prediction accuracy and energy use efficiency of the power consumption monitoring system. It is not only innovative in theory but also demonstrates better energy-saving effects than the existing technologies in practice, providing strong technical support for the field of intelligent energy management.

[0158] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and all of them should be covered by the scope of the claims of the present invention.

Claims

1. A smart electricity monitoring system, characterized in that: include, Data acquisition and preprocessing module: collects environmental parameters, power load status and energy consumption data in real time through environmental sensors and smart meters and performs preprocessing; Multi-dimensional dynamic energy consumption model building module: Integrate the pre-processed data with historical energy consumption records, real-time electricity price information and weather forecast data to build a multi-dimensional dynamic energy consumption model to predict energy consumption trends; Dynamic adjustment strategy formulation module: uses machine learning algorithms to predict energy consumption trends based on multi-dimensional dynamic energy consumption models and formulate dynamic adjustment strategies; Adaptive learning intelligent load balancing module: It uses reinforcement learning technology to dynamically monitor the state of the power grid, dynamically analyze the power grid state based on the predicted energy consumption trend and real-time power grid conditions, and determine the optimal load distribution strategy for each power unit; Strategy execution and control module: receives control instructions from the intelligent load balancing module and automatically executes the load balancing strategy through the intelligent control device; Effect evaluation and feedback optimization module: monitors the actual effect after the implementation of the strategy, collects grid operation status, energy consumption data and user feedback, and evaluates the execution efficiency and energy-saving effect of the strategy; The specific steps of integrating the pre-processed data with historical energy consumption records, real-time electricity price information and weather forecast data are as follows: The pre-processed environmental parameters, power load status and cumulative energy consumption are extracted to form a comprehensive feature vector F, which is integrated with historical energy consumption records, real-time electricity price information and weather forecast data. Based on the integrated data, the fusion feature is constructed, and the expression is: F t =w1·MA k (H t-k:t )+w2·E t +w3·P t +w4·W t +w5·ES α (T t-1:t-k ); Among them, F t represents the comprehensive feature vector at time point t, w1 represents the weight factor of historical energy consumption, t represents the current time point, w2 represents the weight factor of real-time energy consumption, w3 represents the weight factor of real-time electricity price, w4 represents the weight factor of external environmental factors, w5 represents the weight factor of time series trend, MA k (H t-k:t ) represents the sliding average of historical energy consumption at the most recent k time points, E t Represents the real-time energy consumption value, P t represents the real-time electricity price, W t represents the environmental parameter vector, ES α (T t-1:t-k ) represents the exponential smoothing value of the time series trend, α represents the smoothing coefficient, T t-1:t-k Represents the time series data from t-1 to tk; Among them, a composite influence function I is constructed t , the multivariate regression analysis method is used to quantify the impact of each dimension on energy consumption, and the expression is: Among them, I t represents the composite influence function value at time point t, β d Represents the set of regression coefficients in multiple regression analysis, K(x,σ 2 ) represents the Gaussian kernel function, x represents the distance in the feature space, σ 2 represents the variance parameter; Among them, the multi-dimensional impact analysis of energy consumption is combined with fusion features and composite influence functions to build a dynamic energy consumption prediction model and output the dynamic energy consumption prediction value, which is expressed as: Among them, E p (t+h) represents the predicted energy consumption at the time point t+h, h represents the predicted future time span, m represents the order of the piecewise polynomial, c represents the amplitude parameter in the prediction model, T represents the period parameter in the prediction model, represents the sine function term, Norm(.) represents the normalization function, a j and b j Represents the key parameter in polynomial regression.

2. The smart electricity monitoring system according to claim 1, characterized in that: The environmental parameters include temperature, humidity and light; The power load status includes current, voltage, power and accumulated energy consumption.

3. The smart electricity monitoring system according to claim 1, characterized in that: The preprocessing refers to preliminary screening of the collected data, eliminating abnormal values ​​that are obviously beyond the physical range and equipment specifications, and smoothing the data.

4. The smart electricity monitoring system according to claim 1, characterized in that: Using machine learning algorithms, a dynamic adjustment strategy is formulated based on the dynamic energy consumption prediction value. The specific steps are as follows: Analyze the energy consumption forecast curve provided by the dynamic energy consumption forecast value, identify the peak and valley periods of energy consumption, and the key nodes of the energy consumption change rate; Based on historical data, evaluate the energy consumption response sensitivity of each power-consuming device, define the device priority matrix and the minimum power margin for safe operation of the system; Construct the objective function of minimizing the cost of energy consumption and equipment start-up and shutdown costs, and the expression is: Among them, C t represents the minimized cost, N represents the total number of devices, i represents the device index variable, and P c (t) represents the adjusted real-time electricity price, E i (t) represents the real-time energy consumption value of device i at time t, φ i represents the start-stop cost coefficient of equipment i, β represents the coefficient related to the energy consumption cost of equipment adjustment, γ represents the exponential parameter, λ represents the adjustment frequency cost factor, and V(t) represents the frequency of strategy adjustment; Use deep networks to minimize costs and learn the best strategy by interacting with the environment; Deploy the best strategy to the smart power management system, dynamically adjust the operating status and power level of the equipment according to the prediction results, reduce energy consumption in advance before the peak energy consumption, and reasonably utilize resources during the low period.

5. The smart electricity monitoring system according to claim 4, characterized in that: Reinforcement learning technology is used to dynamically monitor the state of the power grid. According to the predicted energy consumption trend and real-time power grid conditions, the power grid state is dynamically analyzed to determine the optimal load distribution strategy for each power unit. The specific operation steps are as follows: Define the state vector S t =[E t ,E p (t+h),P t ,W t ,D t ], where E t Indicates the real-time energy consumption value, W t represents the environmental parameter vector; Create an action space containing all device operation instructions A t ; Design the reward function, the expression is: Where Δf i It represents the deviation of the system frequency from the standard value, and ε represents the scaling factor; Receive state vector S t , generating each possible action a i The expected Q value is expressed as: Among them, R t+1 Indicates taking action A t The immediate feedback obtained at time step t, a′ represents the next state S t+1 Take one of all the actions below and calculate the maximum expected reward for the next state; Execute the selected action and observe the new state S after execution t+1 and instant reward R t , adjust the strategy based on actual grid feedback; The learned strategies are applied to the smart power management system to dynamically adjust the power level and start / stop status of the equipment.

6. The smart electricity monitoring system according to claim 5, characterized in that: Receive control instructions from the intelligent load balancing module and automatically execute the load balancing strategy through the intelligent control device. The specific operation steps are as follows: The monitoring system receives control instructions from the intelligent load balancing module for specific operational requirements of each device; Based on the working status of the current device and the control instructions received, the status is performed; According to the equipment status assessment results, prioritize all equipment to be adjusted, consider energy saving potential, equipment response speed, and factors affecting each other between equipment, and formulate an execution sequence; According to the established execution sequence, precise control signals are sent to each device through the intelligent controller; Collect the real-time data after the equipment is adjusted, compare it with the prediction model, evaluate the effect of strategy execution, and if deviation is found, start the feedback error correction model in time, and make adjustments based on the deviation between the predicted value and the actual value until the expected goal is achieved. The expression is: Among them, ΔP c,i represents the corrected power adjustment of device i, k p Represents the proportional control coefficient, k i Indicates the integral control coefficient, E p,i (t+h) represents the predicted energy consumption of device i at time t+h, E a,i (t+h) represents the actual energy consumption of device i at time t+h, t0 represents the starting time point, and τ represents the integral variable, which represents each time point in the time range from a certain initial time t0 to the current time t.

7. The smart electricity monitoring system according to claim 6, characterized in that: Monitor the actual effect of the strategy after implementation, collect grid operation status, energy consumption data and user feedback, and evaluate the execution efficiency and energy-saving effect of the strategy. The specific steps are as follows: Continuously collect and integrate real-time grid operation status data; A comprehensive evaluation index system including strategy response time, execution accuracy and energy-saving efficiency is established, and the expression is: Among them, ηe represents energy-saving efficiency, Ep_t represents the total energy consumption predicted before the implementation of the strategy, Ea_b represents the actual total energy consumption before the implementation of the energy-saving strategy, and Ea_a represents the actual total energy consumption after the implementation of the energy-saving strategy; Based on time series analysis, combined with the long-term trend and seasonal changes of energy saving, a quantitative model of energy saving effect is constructed, which is expressed as follows: Where ΔEs,v represents the predicted seasonal energy consumption increment at time point v, and B represents the length of the time window for analysis; Design questionnaires or online surveys to collect users’ satisfaction evaluations after the implementation of the strategy and convert them into quantitative indicators; Introduce environmental adaptability indicators to evaluate the adaptability and stability of strategies under different environmental conditions; Combined with the above evaluation results, a feedback optimization model is constructed to dynamically adjust the strategy parameters and equipment operation instructions. The expression is: J=h e +a·U s +b·A e Among them, J represents the maximization of the overall benefit function, Us represents the satisfaction score, and A e represents the environmental adaptability index, a represents the system goal trade-off coefficient, and b represents the user demand trade-off coefficient; According to the results of the feedback optimization model, the strategy parameters are adjusted and redeployed to the smart power management system.

Images