Multi-source integrated energy system energy-saving optimization method based on data mining

Abstract

The embodiment of the invention provides a multi-source integrated energy system energy-saving optimization method based on data mining, and the method comprises the steps: obtaining the real-time operation data of a multi-source integrated energy system, and the real-time operation data comprises the energy demand, supply, environment parameters and equipment state information; utilizing historical operation data and real-time data to construct an energy demand and supply prediction model; analyzing errors of the prediction model by using a data mining technology, and identifying key factors influencing prediction accuracy; adjusting prediction model parameters according to the key factors to improve prediction precision; formulating a scheduling strategy of the multi-source integrated energy system based on the optimized prediction model; and implementing and evaluating the energy-saving effect of the scheduling strategy, and optimizing system operation. Through the scheme of the embodiment of the invention, the problem of prediction accuracy caused by energy demand fluctuation and supply instability in a dynamic operation environment can be solved.

Description

Technical Field

[0001] The present application relates to the field of energy regulation technology, and specifically to a method for energy-saving optimization of a multi-source integrated energy system based on data mining. Background Art

[0002] Energy-saving optimization of multi-source integrated energy systems refers to the process of achieving efficient operation of energy systems and energy conservation and emission reduction goals by integrating multiple energy supply methods (such as electricity, natural gas, renewable energy, etc.) and combining advanced information technology and management strategies. This system can flexibly adjust the energy supply structure according to user needs and changes in the external environment, improve energy utilization efficiency, reduce energy waste, and reduce environmental pollution and carbon emissions, which is of great significance for promoting sustainable development. However, in a dynamic operating environment, this method faces challenges in prediction accuracy due to fluctuations in energy demand and instability in supply. Specifically, energy demand may change suddenly due to factors such as weather, time, and user behavior, while energy supply may also be affected by resource supply and equipment status. These uncertainties make it difficult for prediction models based on historical data to accurately predict future demand and supply, thereby affecting the effect of energy-saving optimization. Summary of the invention

[0003] In view of this, an embodiment of the present disclosure provides a multi-source integrated energy system energy-saving optimization method based on data mining, which at least partially solves the problems existing in the prior art.

[0004] The present invention provides a multi-source integrated energy system energy-saving optimization method based on data mining, comprising: Obtain real-time operating data of multi-source integrated energy systems, including energy demand, supply, environmental parameters and equipment status information; Use historical operating data and real-time data to build a forecasting model for energy demand and supply; Use data mining techniques to analyze the errors of the forecast model and identify the key factors that affect the accuracy of the forecast; Adjust the prediction model parameters according to key factors to improve prediction accuracy; Formulate dispatch strategies for multi-source integrated energy systems based on the optimized forecasting model; Implement and evaluate the energy-saving effect of the scheduling strategy to optimize system operation.

[0005] According to one embodiment, it further comprises: Based on the optimized forecasting model, real-time monitoring of system operation data can be carried out to identify periods of energy demand fluctuations and unstable supply; When energy demand fluctuations are identified, the parameters of the forecasting model are adjusted to reduce the error caused by demand changes; When supply instability is identified, adjust the parameters of the forecast model to improve the stability of the forecast; Update the prediction model regularly to ensure it adapts to changes in the dynamic operating environment.

[0006] According to one embodiment, the real-time monitoring of system operation data based on the optimized prediction model and identifying periods of energy demand fluctuations and unstable supply include: Real-time collection of energy demand and supply data in the system; Calculate the rate of change of energy demand. If the rate of change exceeds the threshold R1, it is considered as demand fluctuation. Calculate the change rate of supply. If the change rate exceeds the threshold R2, it is considered that the supply is unstable. If ΔD>R1 or ΔS>R2, the system enters the early warning state, where ΔD is the rate of change of energy demand, indicating the magnitude of demand change, R1 is the set demand change rate threshold, ΔS is the rate of change of supply, indicating the magnitude of supply change, and R2 is the set supply change rate threshold.

[0007] According to one embodiment, when energy demand fluctuations are identified, adjusting the parameters of the prediction model to reduce errors caused by demand changes includes: Adjust the learning rate α of the prediction model to adapt to demand changes more quickly; Increase the weight λ1 of the forecast model to enhance sensitivity to demand fluctuations; Reduce the smoothing coefficient β to reduce the impact of past data; If ΔD>R1, the adjusted parameters satisfy α = α + ΔD × β, where α is the learning rate, ΔD is the rate of change of energy demand, and β is the smoothing coefficient.

[0008] According to one embodiment, when supply instability is identified, adjusting the parameters of the prediction model to improve the stability of the prediction includes: Reduce the learning rate α of the prediction model to prevent overfitting; Increase the smoothing coefficient β to make the prediction results more stable; Reducing the weight λ1 reduces the sensitivity to instantaneous supply fluctuations; If ΔS>R2, the adjusted parameters satisfy β = β + ΔS × α, where β is the smoothing coefficient, ΔS is the rate of change of supply, and α is the learning rate.

[0009] According to one embodiment, the periodic updating of the prediction model to ensure that it adapts to changes in the dynamic operating environment includes: Collect the latest operation data at fixed intervals T; Retrain the prediction model and update the model parameters; Compare the prediction accuracy of the new and old models and select the model with higher accuracy; If the accuracy improvement ΔP>P1, the prediction model is updated, where ΔP is the improvement in prediction accuracy and P1 is the set accuracy improvement threshold.

[0010] According to one embodiment, it further comprises: Regularly evaluate the performance of the forecast model and calculate the forecast error ε; Adjust the parameters of the prediction model to reduce errors; Save the optimal model parameters for future use; If ε>ε1, the adjusted parameters satisfy θ = θ α × ε, where ε is the prediction error, θ is the model parameter, α is the learning rate, and ε1 is the set error threshold.

[0011] According to one embodiment, it further comprises: Monitor the performance of the forecast model in real time and record historical data of forecast errors; Adjust the threshold parameters R1 and R2 by statistically analyzing historical data; According to the actual operation situation, optimize the threshold setting and improve the accuracy of early warning; If N consecutive monitoring results show that the prediction error ε exceeds ε1, the thresholds R1 = R1 + k1 × ε1 and R2 = R2 + k2 × ε1 are adjusted, where k1 and k2 are adjustment coefficients used to dynamically adjust the thresholds.

[0012] The disclosed embodiment provides a method for energy-saving optimization of a multi-source integrated energy system based on data mining, including: obtaining real-time operating data of the multi-source integrated energy system, including energy demand, supply, environmental parameters and equipment status information; using historical operating data and real-time data to build a prediction model for energy demand and supply; using data mining technology to analyze the errors of the prediction model and identify key factors affecting the accuracy of the prediction; adjusting the prediction model parameters according to the key factors to improve the prediction accuracy; formulating a scheduling strategy for the multi-source integrated energy system based on the optimized prediction model; implementing and evaluating the energy-saving effect of the scheduling strategy to optimize system operation. Through the solution of the disclosed embodiment, the problem of prediction accuracy caused by energy demand fluctuations and supply instability in a dynamic operating environment can be solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the exemplary implementation methods of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the embodiments of the present disclosure and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 It is a flow chart of a method for energy-saving optimization of a multi-source integrated energy system based on data mining; Figure 2 is a flow chart of the energy-saving optimization method of the present application; Figure 3 It is a flowchart for real-time monitoring of system operation data based on the optimized prediction model to identify periods of energy demand fluctuations and unstable supply; Figure 4 It is a flowchart for adjusting the parameters of the forecast model when energy demand fluctuations are identified to reduce the error caused by demand changes; Figure 5 It is a flowchart for adjusting the parameters of the forecast model to improve the stability of the forecast when supply instability is identified; Figure 6 It is a flowchart for regularly updating the prediction model to ensure that it adapts to changes in the dynamic operating environment; Figure 7 is a flow chart of the energy-saving optimization method of the present application; Figure 8 It is a flow chart of the energy-saving optimization method of this application. DETAILED DESCRIPTION

[0015] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0016] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0017] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0018] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0019] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0020] Next, refer to Figure 1 , describes a multi-source integrated energy system energy-saving optimization method based on data mining of the present invention. The method realizes efficient energy-saving optimization of a multi-source integrated energy system (IES) through a series of steps, and solves the problem of prediction accuracy caused by energy demand fluctuations and supply instability in a dynamic operating environment.

[0021] S101: Obtain real-time operating data of the multi-source integrated energy system, including energy demand, supply, environmental parameters and equipment status information. For example, in actual applications, various sensors and technical means can be used to monitor and record in real time the output power of wind and solar energy, temperature changes inside and outside buildings, power load, etc. in a city-level IES. The importance of this step is to ensure that the data obtained can truly reflect the operating status of the system.

[0022] S102: Use historical operation data and real-time data to build a forecast model for energy demand and supply. Specifically, in one embodiment, a support vector machine or long short-term memory network algorithm in machine learning can be used to build a load forecast and renewable energy output estimation model that can better capture the temporal and spatial characteristics, providing a scientific basis for subsequent optimization and scheduling. These models try to predict energy demand and the availability of renewable resources in the future by continuously learning past data patterns and development trends.

[0023] In one embodiment, the prediction model in the present application can be an exponential smoothing model in time series analysis or a long short-term memory network in model deep learning.

[0024] In time series analysis, the exponential smoothing model is a method for predicting future values. It smoothes data by giving different weights to different historical data points to identify the trend of the data. Among them, the learning rate is a parameter between 0 and 1, which indicates the degree of influence of the current observation on the prediction result. When the learning rate is close to 1, the model pays more attention to the most recent data points, so that the predicted value can quickly adapt to new changes; when the learning rate is close to 0, it is more inclined to keep the previous predicted value, making the model less sensitive to short-term fluctuations. This parameter determines the balance between new and old information and controls the speed at which the model responds to changes in the time series. The weight is a value that decays exponentially as the data point moves farther away from the current time point, which means that the latest observations are given higher weights, while earlier data points have less impact. This weight distribution mechanism helps capture trends and seasonal patterns in time series while reducing the impact of noise.

[0025] In the long short-term memory network (LSTM), the learning rate is a hyperparameter that determines the degree to which the model adopts new information during the training process. A higher learning rate can speed up training but may cause model instability or skip the optimal solution. A lower learning rate helps the model converge to the optimal solution more accurately, but may significantly increase the training time. Weights are parameters that connect each neuron in the neural network. They affect the conversion of input data through the forward propagation process and are adjusted according to the gradient of the loss function during the back propagation process to optimize model performance. The smoothing coefficient is usually used in the exponential weighted average calculation when updating weights. It helps the model reduce the fluctuation of weight updates during training, so that the model can learn long-term dependencies more stably.

[0026] S103: Use data mining techniques to analyze the errors of the forecast model and identify the key factors that affect the forecast accuracy. This process aims to identify the main variables that affect the forecast performance so that targeted improvements can be made later. For example, by clustering, classifying and correlating the data of historical forecast errors, researchers may find that certain extreme weather conditions or special holidays are important reasons for the large deviation of the model.

[0027] S104: Adjust the prediction model parameters based on key factors to improve prediction accuracy. This process may involve multiple iterations of experiments until the best configuration that best suits the current situation is found. For example, after knowing that a sudden drop in temperature will affect the prediction results, the prediction accuracy can be improved by adding weather forecast data as model input or retraining a more powerful version of the algorithm.

[0028] S105: Based on the optimized prediction model, formulate a dispatch strategy for the multi-source integrated energy system. The core of this stage is to minimize energy consumption costs and carbon emissions while ensuring the continuity and reliability of power supply. In one embodiment, a set of intelligent optimization algorithms combining energy storage and demand response measures can be designed to store excess renewable energy during low electricity consumption periods and release it during peak hours, which not only avoids electricity waste but also improves energy efficiency.

[0029] S106: Implement and evaluate the energy-saving effect of the scheduling strategy and optimize system operation. By continuously tracking and evaluating system performance, it is possible to verify the energy-saving effect and whether there are new challenges waiting to be solved. Once it is found that the optimization solution in a specific scenario is not ideal, it can be quickly fed back to the previous steps to perform model debugging and strategy optimization again. This iterative process is the key to the energy-saving optimization method proposed in the present invention being able to adapt to the ever-changing dynamic environment and effectively overcome the difficulty of energy prediction.

[0030] Next, refer to Figure 2 , describing the specific steps of formulating a scheduling strategy for a multi-source integrated energy system based on the optimized prediction model of the present invention.

[0031] S201: Real-time monitoring of system operation data to identify periods of energy demand fluctuations and unstable supply. This process involves a data acquisition module that regularly obtains operation data of multiple subsystems such as electricity, heat, and cooling, including but not limited to electricity consumption, heating, cooling capacity, and environmental conditions such as temperature and humidity. By analyzing these data, the peak and trough periods of energy consumption of the system can be identified, thereby providing a basis for subsequent adjustments.

[0032] S202: When the system identifies fluctuations in energy demand, it will adjust the parameters of the prediction model to reduce the error caused by changes in demand. For example, an adaptive learning algorithm, such as gradient descent or random forest algorithm, can be used to automatically adjust the model parameters. Specifically, assume that the prediction model is y = ax + b, where y represents energy demand, x represents influencing factors (such as temperature, time), and a and b are model parameters. When a significant change in demand is detected over a period of time, the prediction error is reduced by updating the values ​​of a and b. These parameters can be fine-tuned in real time according to the actual fluctuations, and the optimal value is usually obtained through cross-validation or grid search methods.

[0033] S203: If unstable supply is identified, the parameters of the prediction model need to be adjusted to improve the stability of the prediction. This step also uses the above-mentioned adaptive learning algorithm, with the focus on reducing the negative impact of supply uncertainty. For example, if the wind power output is unstable during a certain period, the model can adjust the coefficients of wind speed and wind direction on power generation to ensure that energy supply and demand can be accurately predicted even when wind speed varies greatly. The parameter adjustment strategy may focus more on robustness rather than single error minimization.

[0034] S204: Update the forecast model regularly to ensure that it adapts to changes in the dynamic operating environment. During this process, newly generated data is collected and the model is retrained to maintain the accuracy of the model. For example, at the end of each month or quarter, the system will summarize the latest operating data and use this data to retrain the model. This not only helps to capture changes in seasonality and long-term trends, but also adapts to new changes brought about by technological advances. During the entire update process, incremental learning or online learning methods can be used to ensure continuous optimization of the system and improve forecast accuracy.

[0035] Next, refer to Figure 3 , describing the present invention based on the optimized prediction model, real-time monitoring of system operation data, and identification of energy demand fluctuations and supply instability periods.

[0036] S301: Real-time collection of energy demand data and supply data in the system. This process involves obtaining real-time data from multiple sensors and data collection points to ensure the integrity and accuracy of the data. These data include but are not limited to electricity consumption, hot water demand, cooling demand, and the supply of various energy sources.

[0037] S302: Calculate the rate of change of energy demand. The rate of change is obtained by performing a differential operation on the data within a period of time. If the rate of change exceeds the preset threshold R1, it is considered as demand fluctuation. Here, R1 is a value set based on historical data and experience, usually ranging from 0.1 to 0.5, with the optimal value being 0.3. This range is selected because a too low threshold may cause the system to be too sensitive, while a too high threshold may miss important fluctuation information.

[0038] S303: Calculate the change rate of supply. This step is similar to calculating the change rate of demand. If the change rate exceeds the preset threshold R2, the supply is considered unstable. Similarly, R2 is also set based on historical data and experience, generally between 0.1 and 0.5, with the optimal value being 0.4. This setting is intended to ensure that the system can detect and respond to changes in the supply side in a timely manner, so as to take corresponding measures.

[0039] S304: If ΔD>R1 or ΔS>R2, the system enters the early warning state. In this state, the system will trigger an alarm and generate a report to remind operators to pay attention to and deal with potential problems. ΔD represents the magnitude of demand change, R1 is the set demand change rate threshold; ΔS represents the magnitude of supply change, R2 is the set supply change rate threshold. This judgment condition ensures that the system can fully cover fluctuations in both demand and supply, thereby improving the robustness and reliability of the system.

[0040] For example, in a multi-source integrated energy system, the system collects supply and demand data for electricity, water and heat every minute. At a certain moment, the system records that the demand for electricity has increased from 1,000 kW to 1,200 kW in the past five minutes, with a rate of change of 0.4, which exceeds the set threshold of 0.3, and is therefore identified as demand fluctuation. At the same time, the supply of heat energy has dropped from 500 kW to 400 kW in the same period of time, with a rate of change of 0.5, which exceeds the set threshold of 0.4 and is considered to be unstable supply. In this case, the system immediately enters the early warning state and issues an alarm so that the operator can take corresponding measures to ensure the stable operation of the system.

[0041] Next, refer to Figure 4 , describes the present invention, when energy demand fluctuations are identified, adjusting the parameters of the prediction model to reduce the error caused by demand changes. When energy demand fluctuations are identified, in order to better adapt to such changes and reduce prediction errors, it is necessary to dynamically adjust the parameters of the prediction model. The specific steps are as follows: S401: Adjust the learning rate α of the prediction model so that the model can adapt to changes in demand more quickly. The learning rate α determines the speed at which the model accepts new information during the update process. It usually takes a value between 0 and 1, and the optimal value depends on the specific business scenario. Increasing the learning rate can enable the model to adjust to new demand data more quickly, thereby reducing prediction bias. For example, in a multi-source integrated energy system, if a large number of electricity consumption peaks suddenly appear in a short period of time, adjusting the learning rate α from 0.05 to 0.1 can significantly speed up the model's response to this sudden change.

[0042] S402: Increase the weight λ1 of the forecasting model to enhance the model's sensitivity to demand fluctuations. The weight λ1 is used to balance the importance of each input data in the model, and the range is also between 0 and 1. Increasing the weight can make the model pay more attention to current demand fluctuations and improve its responsiveness to short-term demand changes. Specifically, in one embodiment, the weight λ1 is increased from 0.6 to 0.8, which means that in the new forecast, the current data point will be given a higher weight, which helps to quickly adjust the model output to match the new demand state.

[0043] S403: Reduce the smoothing coefficient β to reduce the impact of past data. The role of the smoothing coefficient β is to determine the importance of historical data in the model. The value is between 0 and 1. A smaller β value will make the model more likely to rely on the latest data. By reducing the β value, the impact of past demand fluctuations on the current forecast can be reduced, and the model can pay more attention to recent data changes. For example, adjusting the smoothing coefficient β from 0.3 to 0.1 can reduce the model's reliance on past stable data when facing a sudden increase or decrease in demand, thereby adapting to the current new situation more quickly.

[0044] S404: If the rate of change of energy demand ΔD is greater than the preset threshold R1, the learning rate α is further adjusted, and the specific calculation formula is: α = α + ΔD × β. In this formula, ΔD represents the rate of change of energy demand, and β is the smoothing coefficient. This formula means that when the demand changes greatly, by increasing the learning rate α, the model can adapt to this change more quickly. This setting helps to update the model parameters in time when the demand fluctuates greatly, and reduce the prediction error caused by sudden changes in demand. Assuming R1 is set to 0.2, when the actual demand change rate is 0.3, the new learning rate is calculated according to the formula: α = 0.05+ 0.3 × 0.1 = 0.08, which indicates that the learning rate will be appropriately increased, so that the model converges to the new demand state faster.

[0045] Through the above steps, the present invention can dynamically adjust the parameters of the prediction model when energy demand fluctuations are detected, thereby improving the prediction accuracy and reducing the errors caused by demand changes.

[0046] Next, refer to Figure 5 , describing the present invention, when identifying that the supply is unstable, adjusting the parameters of the prediction model to improve the stability of the prediction. Specifically, in the present application, when the supply of the multi-source integrated energy system is detected to be unstable, the stability of the prediction is improved by adjusting the parameters of the prediction model.

[0047] S501: Reduce the learning rate α of the prediction model to prevent the model from overfitting during training. The learning rate α is a parameter between 0 and 1, usually between 0.001 and 0.1, and the optimal value depends on the specific training data set and problem. Reducing the learning rate can help the model adapt to data changes more slowly, thereby avoiding premature convergence to a local optimal solution.

[0048] S502: Increase the smoothing coefficient β to make the prediction results more stable. The smoothing coefficient β is also between 0 and 1, and the commonly used initial value is about 0.9. Increasing the smoothing coefficient can make the model give higher weight to historical data when predicting, reducing the error caused by short-term fluctuations. For example, in one embodiment, when the system detects that the power supply in the power supply grid fluctuates significantly, the smoothing coefficient β is increased from 0.9 to 0.95, thereby making the prediction results more stable and reducing the inaccuracy caused by short-term fluctuations.

[0049] S503: Reduce the weight λ1 to reduce sensitivity to instantaneous supply fluctuations. The weight λ1 is a parameter in the model used to adjust the importance of each feature, usually between 0 and 1. The optimal value needs to be determined by cross-validation and other methods based on actual conditions. Reducing the value of λ1 can reduce the model's reliance on instant data and make it more focused on long-term trends. In one embodiment, when it is detected that the supply of solar and wind energy fluctuates significantly due to weather changes, λ1 is reduced from 0.7 to 0.5, thereby reducing the model's response to short-term changes.

[0050] S504: If ΔS>R2, the adjusted parameters satisfy β = β + ΔS × α, where β is the smoothing coefficient, ΔS is the rate of change of supply, and α is the learning rate. This formula means that when the rate of change of supply exceeds a certain threshold R2, the smoothing coefficient β is further increased to enhance the smoothing effect. For example, let ΔS be the rate of change of supply and R2 be the predetermined threshold 0.1. If ΔS is detected to be 0.15, it is adjusted according to the formula β = β + 0.15 × α. Assuming that the current smoothing coefficient β is 0.9 and the learning rate α is 0.05, the adjusted β is 0.9 + 0.15 × 0.05 = 0.9075. Doing so can better cope with larger supply fluctuations and ensure that the output of the forecasting model is more reliable and stable. In this way, when supply instability is identified, the parameters of the forecasting model can be effectively adjusted to improve the stability of its forecast.

[0051] Next, refer to Figure 6 , describing the periodic updating prediction model of the present invention to ensure that it adapts to changes in the dynamic operating environment.

[0052] S601: Collect the latest operating data at fixed intervals T. This means that the system automatically obtains new data points at specific time intervals, which may include various operating indicators such as temperature, energy consumption, power generation, etc. This is done to ensure that the prediction model is always updated based on the latest and most accurate data to improve the model's prediction ability. For example, in a multi-source integrated energy system, the system collects new operating data from each energy supply unit, sensor, and device every 24 hours.

[0053] S602: Retrain the prediction model and update the model parameters. This process involves retraining the existing model using the newly collected data to update the internal parameters of the model. The purpose of the update is to enable the model to better fit the new data distribution and improve the accuracy of the model in the current operating environment. Specifically, in a multi-source integrated energy system, the system uses machine learning algorithms (such as support vector machines, neural networks, etc.) and newly collected operating data to retrain the energy-saving optimization model, thereby adjusting the model's parameters such as weights and biases.

[0054] S603: Compare the prediction accuracy of the new and old models and select the model with higher accuracy. After updating the model, the system will compare the prediction accuracy of the new model and the old model to ensure that the new model performs better than or at least not worse than the old model in actual applications. Prediction accuracy can usually be evaluated by indicators such as mean square error (MSE), mean absolute error (MAE) or determination coefficient (R²). For example, in a specific embodiment of the energy-saving optimization method for a multi-source integrated energy system, the system calculates the MAE of the new and old models on the same test set, and selects the model with a lower MAE as the final optimization model.

[0055] S604: If the accuracy improvement ΔP>P1, update the prediction model, where ΔP is the improvement in prediction accuracy and P1 is the set accuracy improvement threshold. ΔP represents the accuracy difference between the new model and the old model, and P1 is a pre-set threshold used to determine whether it is necessary to update the model. The setting of P1 is usually based on the requirements of actual applications. Too high a P1 may lead to frequent but unnecessary model updates, while too low a P1 may ignore significant performance improvements. Usually, the value range of P1 is between 0 and 1, and the optimal value depends on the application scenario. For example, in some energy-saving optimization tasks of multi-source integrated energy systems, if P1 is set to 0.05, it means that the prediction accuracy of the new model needs to be improved by at least 5% compared to the old model before the model is updated.

[0056] This series of steps ensures that the prediction model can continuously adapt to the dynamically changing operating environment, thereby maintaining high prediction accuracy and system performance.

[0057] Next, if Figure 7 As shown, another embodiment of the energy system energy saving optimization method of the present invention is described. The method of this embodiment includes: S701: Regularly evaluate the performance of the prediction model and calculate the prediction error ε. This step determines the accuracy of the model by comparing the model prediction results with the actual data. Specifically, the difference between the predicted value and the actual value is calculated to form an error index. For example, in a practical application, the prediction error ε can be obtained by comparing the energy consumption predicted by the model with the energy consumption actually recorded by the system.

[0058] S702: According to the calculated prediction error, adjust the parameters of the prediction model to reduce the error. This is achieved through back propagation or other optimization algorithms, and the purpose is to make the model more accurately reflect the actual situation. In one embodiment, when the energy consumption predicted by the model deviates significantly from the actual value, certain weight parameters in the model can be increased or decreased to reduce the prediction error.

[0059] S703: Save the optimal model parameters for future use. This step is to ensure that the model can still maintain a high accuracy in future applications. Specifically, if the error is reduced after each adjustment, the current model parameters are saved to avoid repeated calculations. In an application scenario, the system will automatically store these optimal parameters in the database for direct call next time.

[0060] S704: If the calculated prediction error ε exceeds the set error threshold ε1, the model parameters are further adjusted to make them more optimized. The adjustment is performed through the formula θ = θ + α × ε, where θ represents the model parameters, α represents the learning rate, and ε represents the prediction error. The learning rate α determines the speed of parameter adjustment and is usually set to a decimal between 0 and 1. The formula means that the prediction error is multiplied by the learning rate α and added to the current parameter θ, so that the model gradually approaches the optimal parameter. In a specific example, assuming that the learning rate α is set to 0.01, if the prediction error of the model is large, the system will automatically use the above formula to update the model parameters until the prediction error falls within an acceptable range. The formula is set in this way to allow the model to gradually converge in multiple iterations, improve the prediction accuracy, and ensure efficient and energy-saving operation of the energy system.

[0061] Next, refer to Figure 8 , describes another embodiment of the energy-saving optimization method of the present invention, the method specifically comprising: S801: Monitor the performance of the prediction model in real time and record historical data of prediction errors; S802: adjusting threshold parameters R1 and R2 by statistically analyzing historical data; S803: Optimize threshold settings according to actual operation conditions to improve the accuracy of early warning; S804: If N consecutive monitoring results show that the prediction error ε exceeds ε1, the thresholds R1 = R1 + k1× ε1 and R2 = R2 + k2 × ε1 are adjusted, where k1 and k2 are adjustment coefficients.

[0062] Specifically, the first step is to monitor the performance of the prediction model in real time and record the historical data of the prediction error. This means that through sensors and data acquisition equipment, the system's operating data in different time periods, including power consumption, heat use, ambient temperature, etc., is continuously collected, compared with the data predicted by the model, and the actual deviation is recorded each time. For example, in a multi-source integrated energy system, the difference between power generation and power demand can be monitored in real time, and each prediction error can be recorded in the database for subsequent analysis.

[0063] Next, the threshold parameters R1 and R2 are adjusted by statistically analyzing these historical data. This process involves data mining techniques, such as regression analysis or cluster analysis, which are used to discover the law of error distribution. Based on the analysis results, it can be determined whether the current threshold is appropriate and adjusted if necessary. For example, by analyzing the forecast errors of the past year, it is found that the errors under certain specific conditions are too large, and the sensitivity of the warning can be increased by reducing R1.

[0064] Then, according to the actual operation situation, the threshold setting is optimized to improve the accuracy of early warning. This step relies on feedback from actual operation. By continuously testing different threshold combinations, the optimal threshold setting is selected to improve the early warning accuracy and response speed of the system. For example, in one embodiment, through verification of multiple actual operation data, it is found that setting R1 to 10% can more accurately capture energy consumption anomalies, while setting R2 to 20% can avoid frequent false alarms.

[0065] Finally, if N consecutive monitoring results show that the prediction error ε exceeds ε1, the thresholds R1 = R1 + k1× ε1 and R2 = R2 + k2 × ε1 are adjusted, where k1 and k2 are adjustment coefficients. Here, N represents the number of consecutive monitoring times, and ε1 is a preset error tolerance value used to determine whether the error exceeds the acceptable range. The adjustment coefficients k1 and k2 determine the adjustment range of R1 and R2 respectively. For example, assuming N=5, ε1=15%, k1=0.5, k2=0.3, when the prediction error exceeds 15% for 5 consecutive times, the system will automatically increase R1 by 0.5 × 15% = 7.5% and R2 by 0.3 × 15% = 4.5%. Such adjustments can ensure that the system responds quickly to emergencies and reduce losses caused by model errors.

[0066] In summary, this method significantly improves the energy-saving optimization effect and early warning accuracy of the multi-source integrated energy system through real-time monitoring, data analysis and dynamic adjustment of thresholds.

[0067] A multi-source integrated energy system energy-saving optimization method based on data mining of the present invention first obtains real-time operation data from the multi-source integrated energy system, which includes but is not limited to energy demand, energy supply, environmental parameters and equipment status information. These data are the basis for subsequent steps and ensure a comprehensive understanding of the system operation status. Then, by combining historical operation data with real-time data, a model for predicting energy demand and supply is constructed. The purpose of this step is to establish a model that can effectively predict future energy demand and supply by learning a large amount of historical data, so as to provide a basis for the optimized operation of the system. Then, the data mining technology is applied to perform error analysis on the constructed prediction model to identify the key factors affecting the prediction accuracy. This step is crucial because systems in the real world are often affected by various uncertainties and complex factors, and identifying these key factors helps to more accurately tune the prediction model. Then, according to the key factors identified in the previous step, the parameters of the prediction model are adjusted. This process is a continuous iterative process, which aims to gradually improve the prediction accuracy of the model, make it closer to the actual situation, and reduce the prediction error. After obtaining a high-precision prediction model, the present invention then formulates a scheduling strategy for the multi-source integrated energy system based on the model. Through this strategy, we can achieve the best resource allocation and minimize energy consumption while ensuring the energy needs of users. Finally, put the above-developed scheduling strategy into practice, and monitor and evaluate its actual energy-saving effect. If it is found that the effect does not meet expectations or there is room for further optimization, the previous steps can be repeated to continuously adjust and improve the entire energy-saving optimization process.

[0068] Through this series of steps, the present invention effectively solves the problem of prediction accuracy caused by fluctuations in energy demand and unstable supply in a dynamic operating environment, thereby improving the overall energy efficiency of the multi-source integrated energy system and reducing energy consumption costs.

[0069] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0070] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0071] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0072] Any process or method description in the flow chart or otherwise described herein can be understood to represent a module, fragment or portion of a code including one or more executable instructions for implementing the steps of a specific logical function or process. And the scope of the preferred embodiment of the present application includes other implementations, in which the functions may not be performed in the order shown or discussed, including in a substantially simultaneous manner or in a reverse order according to the functions involved.

[0073] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with these instruction execution systems, apparatuses or devices.

[0074] It should be understood that the various parts of the present application can be implemented with hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented with software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above embodiment method can be completed by instructing the relevant hardware through a program, which can be stored in a computer-readable storage medium, and when the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0075] In addition, each functional unit in each embodiment of the present application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the above-mentioned integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium can be a read-only memory, a disk or an optical disk, etc.

[0076] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A multi-source integrated energy system energy-saving optimization method based on data mining, characterized in that: include: Obtain real-time operating data of multi-source integrated energy systems, including energy demand, supply, environmental parameters and equipment status information; Use historical operating data and real-time data to build a forecasting model for energy demand and supply; Use data mining techniques to analyze the errors of the forecast model and identify the key factors that affect the accuracy of the forecast; Adjust the prediction model parameters according to key factors to improve prediction accuracy; Formulate dispatch strategies for multi-source integrated energy systems based on the optimized forecasting model; Implement and evaluate the energy-saving effect of the scheduling strategy to optimize system operation.

2. According to claim 1, a multi-source integrated energy system energy-saving optimization method based on data mining is characterized in that: Also includes: Based on the optimized forecasting model, real-time monitoring of system operation data can be carried out to identify periods of energy demand fluctuations and unstable supply; When energy demand fluctuations are identified, the parameters of the forecasting model are adjusted to reduce the error caused by demand changes; When supply instability is identified, adjust the parameters of the forecast model to improve the stability of the forecast; Update the prediction model regularly to ensure it adapts to changes in the dynamic operating environment.

3. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 2 is characterized in that: The above-mentioned real-time monitoring of system operation data based on the optimized prediction model and identification of energy demand fluctuations and supply instability periods include: Real-time collection of energy demand and supply data in the system; Calculate the rate of change of energy demand. If the rate of change exceeds the threshold R1, it is considered as demand fluctuation. Calculate the change rate of supply. If the change rate exceeds the threshold R2, it is considered that the supply is unstable. If ΔD > R1 or ΔS > R2, the system enters the early warning state, where ΔD is the rate of change of energy demand, indicating the magnitude of demand change, R1 is the set demand change rate threshold, ΔS is the rate of change of supply, indicating the magnitude of supply change, and R2 is the set supply change rate threshold.

4. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 3 is characterized in that: When energy demand fluctuations are identified, the parameters of the prediction model are adjusted to reduce the error caused by demand changes, including: Adjust the learning rate α of the prediction model to adapt to demand changes more quickly; Increase the weight λ1 of the forecast model to enhance sensitivity to demand fluctuations; Reduce the smoothing coefficient β to reduce the impact of past data; If ΔD > R1, the adjusted parameters satisfy α = α + ΔD × β, where α is the learning rate, ΔD is the rate of change of energy demand, and β is the smoothing coefficient.

5. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 3 is characterized in that: When supply instability is identified, adjusting the parameters of the forecast model to improve the forecast stability includes: Reduce the learning rate α of the prediction model to prevent overfitting; Increase the smoothing coefficient β to make the prediction results more stable; Reducing the weight λ1 reduces the sensitivity to instantaneous supply fluctuations; If ΔS > R2, the adjusted parameters satisfy β = β + ΔS × α, where β is the smoothing coefficient, ΔS is the rate of change of supply, and α is the learning rate.

6. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 2 is characterized in that: The periodic updating of the prediction model to ensure that it adapts to changes in the dynamic operating environment includes: Collect the latest operation data at fixed intervals T; Retrain the prediction model and update the model parameters; Compare the prediction accuracy of the new and old models and select the model with higher accuracy; If the accuracy improvement ΔP > P1, the prediction model is updated, where ΔP is the improvement in prediction accuracy and P1 is the set accuracy improvement threshold.

7. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 3 is characterized in that: Also includes: Regularly evaluate the performance of the forecast model and calculate the forecast error ε; Adjust the parameters of the prediction model to reduce errors; Save the optimal model parameters for future use; If ε > ε1, the adjusted parameters satisfy θ = θ+α × ε, where ε is the prediction error, θ is the model parameter, α is the learning rate, and ε1 is the set error threshold.

8. The method for energy saving optimization of a multi-source integrated energy system based on data mining according to claim 7 is characterized in that: Also includes: Monitor the performance of the forecast model in real time and record historical data of forecast errors; Adjust the threshold parameters R1 and R2 by statistically analyzing historical data; According to the actual operation situation, optimize the threshold setting and improve the accuracy of early warning; If N consecutive monitoring results show that the prediction error ε exceeds ε1, the thresholds R1 = R1 + k1 × ε1 and R2 = R2 + k2 × ε1 are adjusted, where k1 and k2 are adjustment coefficients used to dynamically adjust the thresholds.

Images