Thermoelectric load prediction device and method combined with demand side energy consumption analysis

Abstract

The invention provides a thermoelectric load prediction device and method in combination with demand side energy consumption analysis, and relates to the technical field of power system load prediction. The data preprocessing module preprocesses the thermoelectric load data set to obtain a standard thermoelectric load data set; the data analysis module performs demand side energy consumption analysis and influence factor extraction on the standard thermoelectric load data set to obtain a thermoelectric load influence factor set; a data feature extraction module performs feature extraction on the standard thermoelectric load data set to obtain a thermoelectric load feature set; and the load prediction module constructs a deep learning model list to carry out training optimization on the thermoelectric load feature set, and obtains a thermoelectric load adaptive prediction model to carry out thermoelectric load prediction. The technical problem of insufficient load prediction result accuracy caused by lack of deep analysis of terminal demand side energy consumption data in thermoelectric load prediction in the prior art is solved, and the accuracy and reliability of thermoelectric load prediction are improved.

Description

Technical Field

[0001] The present application relates to the technical field of power system load forecasting, and specifically to a thermal power load forecasting device and method combined with demand-side energy analysis. Background Art

[0002] As a renewable energy utilization method, the biomass and solar energy complementary cogeneration system has attracted widespread attention in recent years. This cogeneration system combines the advantages of biomass energy and solar energy, can achieve clean and efficient cogeneration, and is of great significance in reducing energy consumption and carbon emissions.

[0003] At present, in the thermal power load forecasting methods for biomass and solar complementary cogeneration systems, load forecasting usually focuses on historical load data or power generation data within the system, and rarely considers the specific energy consumption patterns and demand fluctuations on the demand side (i.e., end users). The energy consumption data on the demand side includes a variety of factors such as users' electricity consumption habits, lifestyles, seasonal changes, holiday effects, etc., all of which will have an impact on thermal power loads. Existing forecasting methods ignore the in-depth analysis of these demand-side energy consumption data, resulting in the inability of the forecasting model to fully capture the fluctuations and influencing factors of load demand, reducing the accuracy of the forecast results. Summary of the invention

[0004] The present application provides a thermal power load prediction device and method combined with demand-side energy consumption analysis, which solves the technical problem that the thermal power load prediction in the prior art lacks in-depth analysis of terminal demand-side energy consumption data, resulting in insufficient accuracy of load prediction results, and achieves the technical effect of improving the accuracy and reliability of thermal power load prediction.

[0005] In view of the above problems, on the one hand, the present application provides a thermal power load prediction device combined with demand-side energy consumption analysis, the device comprising: a data acquisition module, used to collect and obtain a thermal power load data set of a biomass and solar energy complementary cogeneration system, the thermal power load data set comprising historical load data and terminal demand-side energy consumption data; a data preprocessing module, used to preprocess the thermal power load data set to obtain a standard thermal power load data set; a data analysis module, used to perform demand-side energy consumption analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set; a data feature extraction module, used to perform feature extraction on the standard thermal power load data set based on the thermal power load influencing factor set to obtain a thermal power load feature set; a load prediction module, used to construct a deep learning model list, use the deep learning model list to train and optimize the thermal power load feature set, obtain a thermal power load adaptive prediction model, and perform thermal power load prediction based on the thermal power load adaptive prediction model.

[0006] On the other hand, the present application also provides a thermal power load prediction method combined with demand-side energy consumption analysis, the method comprising: collecting and obtaining a thermal power load data set of a biomass and solar energy complementary cogeneration system, the thermal power load data set comprising historical load data and terminal demand-side energy consumption data; preprocessing the thermal power load data set to obtain a standard thermal power load data set; performing demand-side energy consumption analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set; performing feature extraction on the standard thermal power load data set based on the thermal power load influencing factor set to obtain a thermal power load feature set; constructing a deep learning model list, using the deep learning model list to train and optimize the thermal power load feature set to obtain a thermal power load adaptive prediction model, and performing thermal power load prediction based on the thermal power load adaptive prediction model.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The data acquisition module acquires the thermal power load data set of the biomass and solar energy complementary cogeneration system. The thermal power load data set includes historical load data and terminal demand side energy consumption data. The historical load data can reflect the past operation of the system and provide a time series reference for load forecasting. The terminal demand side energy consumption data can reflect the actual energy demand of the user. These thermal power load data sets provide raw data for subsequent analysis, processing and forecasting. The data preprocessing module preprocesses the thermal power load data set to obtain a standard thermal power load data set to improve the quality of the data, making the subsequent analysis and processing more accurate and reliable. The data analysis module performs demand side energy consumption analysis and influencing factor extraction on the standard thermal power load data set to find out the key factors affecting the thermal power load, such as weather, user behavior, etc., and obtains a thermal power load influencing factor set to provide an important basis for subsequent feature extraction and model construction. The data feature extraction module extracts features from the standard thermal power load data set based on the thermal power load influencing factor set to obtain a thermal power load feature set, thereby reducing data redundancy, making the data more representative, and providing more targeted data input for building an efficient and accurate prediction model. The load prediction module constructs a deep learning model list, uses the deep learning model list to train and optimize the thermal power load feature set, obtains a thermal power load adaptive prediction model, and performs thermal power load prediction based on the thermal power load adaptive prediction model to enhance the accuracy and adaptability of load prediction.

[0009] In summary, this application introduces demand-side energy analysis and influencing factor extraction to construct a thermal power load adaptive prediction model that can more comprehensively capture the multi-dimensional load change rules, better adapt to actual energy demand changes, and provide support for the stable operation of the biomass and solar complementary cogeneration system. The adaptability of the deep learning model enables the solution to optimize the prediction results in real time and adapt to load demand fluctuations in complex dynamic environments, thereby improving the accuracy and reliability of thermal power load prediction, providing a scientific basis for efficient energy scheduling of the cogeneration system, and further optimizing energy utilization and load management.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of the structure of a thermal power load prediction device combined with demand-side energy analysis provided in an embodiment of the present application.

[0012] Figure 2 A flow chart of a thermal power load prediction method combined with demand-side energy analysis provided in an embodiment of the present application.

[0013] Explanation of the accompanying drawings: data acquisition module 10, data preprocessing module 20, data analysis module 30, data feature extraction module 40, load prediction module 50. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a thermal power load prediction device and method combined with demand-side energy consumption analysis, thereby solving the technical problem that the thermal power load prediction in the prior art lacks in-depth analysis of terminal demand-side energy consumption data, resulting in insufficient accuracy of load prediction results, and achieves the technical effect of improving the accuracy and reliability of thermal power load prediction.

[0015] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a thermal power load prediction device combined with demand-side energy analysis, the device comprising:

[0016] The data collection module 10 is used to collect and obtain a thermal load data set of the biomass and solar energy complementary cogeneration system, wherein the thermal load data set includes historical load data and energy consumption data on the terminal demand side.

[0017] Specifically, the thermal power load data set is a set of data about the load conditions in the combined heat and power system, including historical load data and terminal demand-side energy consumption data. Among them, the historical load data is a record of the load borne by the combined heat and power system in the past period of time, such as the electricity and heat demand data during the peak and trough periods of each month in the past year; the terminal demand-side energy consumption data is the energy consumption data from the perspective of the end-user who uses energy, such as the electricity and heat consumption habits of household users in different seasons and time periods.

[0018] Electricity sensors can be installed on power generation equipment and transmission lines of the biomass and solar energy complementary cogeneration system, and heat sensors can be installed on heating pipes to monitor the generation and transmission of electricity and heat in real time, thereby obtaining historical load data. For energy consumption data on the terminal demand side, smart electricity meters and smart heat meters can be used to record the electricity and heat consumption of household or corporate users, and user survey data can also be used to understand user usage habits.

[0019] Collect historical load data and terminal demand-side energy consumption data of the biomass and solar complementary heat and power cogeneration system, summarize them in time series to generate a heat and power load data set, and provide the original data source for subsequent analysis, processing and prediction. Historical load data can reflect the past operation of the system and provide a time series reference for prediction, while terminal demand-side energy consumption data can reflect the actual energy demand of the user, such as the user's demand for heat and electricity in different time periods.

[0020] The data preprocessing module 20 is used to preprocess the thermal and electric load data set to obtain a standard thermal and electric load data set.

[0021] Specifically, the collected thermal load data is preprocessed, including removing invalid data, filling missing values, removing noise, and converting all data into a consistent standard format, and finally obtaining a standard thermal load data set. By preprocessing the thermal load data set to obtain a standard thermal load data set, the quality of the data can be improved, making subsequent analysis and processing more accurate and reliable.

[0022] The data analysis module 30 is used to perform demand-side energy analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set.

[0023] Specifically, the thermal power load influencing factor set is a collection of all factors that affect the thermal power load, including weather factors such as temperature and sunshine duration, as well as user behavior factors.

[0024] Use statistical analysis or machine learning methods to mine standard data sets and identify key factors that affect load demand. For example, time periods (daytime and nighttime), seasons (heating in winter, cooling in summer), and meteorological data (such as temperature and humidity), etc. Integrate these influencing factors to obtain a set of thermal power load influencing factors, determine the driving factors of load changes, and thus provide a more comprehensive input for the prediction model.

[0025] The data feature extraction module 40 is used to extract features from the standard thermal power load data set based on the thermal power load influencing factor set to obtain a thermal power load feature set.

[0026] Specifically, the thermal power load feature set is a representative feature data set extracted from the standard thermal power load data set based on the thermal power load influencing factor set. The features in this set can reflect the key characteristics of the thermal power load, such as the average daily power consumption and the peak heat demand in different seasons.

[0027] Based on the set of thermal power load influencing factors, relevant features are extracted from the standard thermal power load data set. For example, the thermal power load influencing factors include temperature, sunshine duration, user electricity usage time and other factors. Through feature extraction, the features that best represent the thermal power load changes are found, and these extracted features are integrated to generate a thermal power load feature set, thereby reducing data redundancy and making the data more representative, providing more targeted data input for building an efficient and accurate prediction model.

[0028] The load prediction module 50 is used to construct a deep learning model list, use the deep learning model list to train and optimize the thermal power load feature set, obtain a thermal power load adaptive prediction model, and perform thermal power load prediction based on the thermal power load adaptive prediction model.

[0029] Specifically, the deep learning model list is a collection of deep learning models. The thermal load adaptive prediction model is a model obtained by training and optimizing the thermal load feature set through the deep learning model list. This model can automatically adjust its own parameters according to new data and changes, so as to better adapt to different thermal load prediction needs. For example, when the seasons change or user habits change, this model can automatically adjust the prediction strategy to maintain a high prediction accuracy.

[0030] A list of deep learning models is constructed, and deep learning models such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are included in them. Then, this list of deep learning models is used to train and optimize the thermal power load feature set. During the training process, the thermal power load feature set is divided into a training set, a validation set, and a test set. The model learns the patterns in the data on the training set, adjusts the model's hyperparameters (such as the number of neural network layers, the number of neurons in each layer, etc.) on the validation set, and finally evaluates the performance of the model on the test set. After multiple iterations of training, until the model converges, the trained deep learning model is output as the thermal power load adaptive prediction model. When predicting the thermal power load, the new thermal power load feature data is input into the thermal power load adaptive prediction model, and the model will output the predicted thermal power load results. For example, the temperature forecast data for a period of time in the future, the data on changes in users' electricity consumption habits, etc. are converted into thermal power load feature data and input into the model, and the model can predict the thermal power load situation in the future period of time.

[0031] Furthermore, the data preprocessing module 20 in the embodiment of the present application is also used to perform the following steps:

[0032] Step P21: normalizing and time-aligning the thermal and electric load data set to obtain a thermal and electric load sequence data set.

[0033] Step P22: performing abnormal data identification on the thermal and electric load sequence data set to obtain abnormal thermal and electric load data, wherein the abnormal thermal and electric load data includes inconsistent data, missing values, and out-of-range values.

[0034] Step P23: pre-process and analyze the abnormal thermal and electric load data to determine the data pre-processing steps.

[0035] Step P24: preprocessing the abnormal thermal and electric load data based on the data preprocessing step to obtain the standard thermal and electric load data set.

[0036] Specifically, each type of data in the thermal power load data set is normalized. The minimum-maximum normalization method can be used to process each type of data separately to eliminate the impact of different dimensions and scales on data analysis and modeling. Normalization can make each data point on the same scale to avoid certain features dominating model training due to a large numerical range. The normalized thermal power load data set is time-aligned, and data from multiple sources are arranged at the same time interval to obtain a thermal power load sequence data set, so that subsequent analysis can be based on a unified time node.

[0037] Statistical analysis methods (such as standard deviation method, box plot method, regression analysis, etc.) are used to find out the data in the thermal load sequence data set that do not conform to normal rules or expectations, and mark them as abnormal thermal load data. These abnormal thermal load data may be data that are logically or numerically inconsistent with adjacent data points, or they may be missing items in the data set, that is, vacant values, or values ​​that exceed the normal or reasonable range, that is, out-of-range values.

[0038] According to the abnormal type of abnormal thermal and electric load data, select the appropriate processing method. For inconsistent data, it is necessary to correct it through other data sources; for vacant values, the mean filling or forward filling method can be used; and out-of-range values ​​may need to be smoothed or deleted. Based on the types of all abnormal thermal and electric load data, determine the steps for preprocessing the entire thermal and electric load sequence data set. Then, the abnormal thermal and electric load data is processed according to the determined data preprocessing steps to generate a standard thermal and electric load data set, providing high-quality input data for subsequent analysis, feature extraction, and prediction.

[0039] Furthermore, the data feature extraction module 40 of the embodiment of the present application is also used to perform the following steps:

[0040] Step P31: Extract features of the standard thermal power load data set based on the thermal power load influencing factor set to obtain a load-related influencing feature set.

[0041] Step P32: Evaluate the correlation degree of each influencing feature in the load correlation influencing feature set to obtain a load influencing feature correlation degree set.

[0042] Step P33: Set the feature correlation threshold according to the thermal power load forecast demand information.

[0043] Step P34: Screening the load impact feature correlation set based on the feature correlation threshold to obtain the thermal power load feature set.

[0044] Specifically, based on the set of thermal power load influencing factors, feature engineering methods are used to extract features from the standard thermal power load data set. For example, statistical methods can be used to calculate the statistical characteristics between each factor and the thermal power load. If the correlation coefficient is high, the factor is extracted. For user behavior factors, features such as the proportion of electricity consumption in different time periods can be counted and extracted. The extracted load features are summarized to determine the load correlation influence feature set. This set contains all features related to load changes, and each feature has a certain correlation with the fluctuation of load demand.

[0045] For each feature in the load association influence feature set, evaluate its degree of association with the thermal power load through mathematical or statistical methods. The correlation coefficient, information entropy, etc. can be used to evaluate the correlation between the feature and the load. For example, for numerical features, the Pearson correlation coefficient is used to measure the degree of association between the feature and the thermal power load; the mutual information method can also be used to evaluate the degree of association between the features. Traverse all the features in the load association influence feature set to evaluate the degree of association, and the obtained values ​​of the degree of association of each feature are combined into a load influence feature association set, which reflects the degree of influence of each influencing feature on the thermal power load.

[0046] According to the accuracy requirements of thermal power load forecasting, the time range of forecasting (short-term, medium-term or long-term forecasting), the targeted user groups or regions, etc., a feature correlation threshold is set to screen the important features in the feature set. For example, based on experience or previous experimental analysis, if a high-precision short-term forecast is to be achieved, the correlation needs to be above 0.8, so the feature correlation threshold for short-term forecasting with high-precision forecasting requirements is set to 0.8. If it is a long-term forecast with relatively low accuracy requirements, a lower threshold is set, such as 0.5.

[0047] Traverse each feature correlation value in the load impact feature correlation set and compare it with the feature correlation threshold. If the feature correlation value is greater than or equal to the feature correlation threshold, the corresponding impact feature will be retained; if the feature correlation value is less than the feature correlation threshold, the corresponding impact feature will be excluded. The features retained after screening constitute the thermal power load feature set. The screened feature set contains features with strong predictive power for load forecasting, avoiding inefficient or irrelevant features from interfering with model training.

[0048] Furthermore, the load prediction module 50 of the embodiment of the present application is also used to perform the following steps:

[0049] Step P51: Perform data characteristic analysis on the thermal power load characteristic set to obtain data characteristic information.

[0050] Step P52: Determine data prediction demand characteristic information based on the thermal power load prediction demand information and the data characteristic information.

[0051] Step P53: Based on the data prediction demand characteristic information, match and optimize with the deep learning model list to determine the target deep learning model.

[0052] Step P54: Use the target deep learning model to train and optimize the thermal power load feature set to obtain the thermal power load adaptive prediction model.

[0053] Specifically, data characteristic information includes data distribution, trend, volatility and other characteristics, which help understand the structure and pattern of data. Data characteristic analysis is performed on the thermal power load feature set to calculate the mean, variance, correlation, periodicity, etc. of each characteristic data, and clarify the data pattern and regularity of these characteristics. For example, the mean and variance of load characteristics are calculated by statistical methods to evaluate the volatility of load demand; or the correlation between certain characteristics (such as temperature) and load demand is analyzed to determine the role of these characteristics in load forecasting.

[0054] Based on the thermal power load forecasting demand information (such as forecasting accuracy, response time, etc.) and the data characteristic information, the key characteristics that need to be paid attention to in the data forecasting process are further extracted, namely the data forecasting demand characteristic information. For example, for real-time forecasting demand, more attention may be paid to short-term load fluctuations; for long-term forecasting, more attention may be paid to trends and seasonal changes.

[0055] By matching the data prediction demand characteristics information with each model in the deep learning model list, the target model that best suits the current task is selected. For example, if the thermal load data has a strong temporal nature, it is more suitable for the long short-term memory network model because it can capture long-term dependencies; if the thermal load data contains spatial features, the convolutional neural network may be more effective. By matching the optimal model, it is ensured that the selected model best meets the requirements of the prediction task, thereby improving the accuracy of load prediction.

[0056] The selected target deep learning model is used to train and optimize the thermal power load feature set. The model parameters are optimized through the back propagation algorithm and gradient descent method to reduce the prediction error and improve the accuracy. For example, if the long short-term memory network model is used, the training process will enable the model to better predict the time series trend of load changes by continuously adjusting the network weights. Ultimately, the trained model will be able to accurately predict thermal power load based on real-time input data.

[0057] The above steps match the appropriate model for training according to the specific requirements and data characteristics of load forecasting, ensuring the high accuracy and reliability of load forecasting, thereby providing real-time and accurate load forecasting in practical applications and improving the efficiency of energy management and resource scheduling.

[0058] Furthermore, step P54 also includes:

[0059] Step P541: Divide the thermal power load feature set into data proportions to obtain a data training set, a data verification set, and a data test set.

[0060] Step P542: Use the target deep learning model to perform feature recognition training on the data training set to generate a basic thermal power load prediction model.

[0061] Step P543: Use the data verification set and the data test set to verify and optimize the basic thermal power load prediction model to obtain the thermal power load adaptive prediction model.

[0062] Specifically, the thermal load feature set is divided into training set, validation set and test set according to a certain ratio. The common division ratio is 70% training set, 15% validation set and 15% test set. The training set is used for model learning, the validation set is used to tune the model parameters, and the test set is used to evaluate the final performance of the model. Reasonable data division can ensure that the model has sufficient data for training and evaluation at different stages, thereby improving the prediction performance of the model.

[0063] Input the input features (such as temperature, user electricity usage habits, and other features that affect thermal power load) and the corresponding output labels (thermal power load values) in the data training set into the target deep learning model. For the target deep learning model, if it is a recurrent neural network or a long short-term memory network, the data will be input into the model in the order of the time series. The target deep learning model is trained using a deep learning framework so that the model can identify the relationship between thermal power load changes and input features, and the trained model is output as the basic thermal power load prediction model.

[0064] Use the validation set to check the performance of the model. If the model overfits on the validation set (i.e., the training accuracy is high but the validation accuracy is low), the network structure or regularization parameters may need to be adjusted. Then, use the test set for final evaluation to ensure that the model has good generalization ability. For example, by adjusting the learning rate, batch size, or optimizer type, the prediction effect of the model on unseen data can be gradually improved, and an accurate and stable thermal power load adaptive prediction model can be finally obtained.

[0065] Furthermore, step P543 also includes:

[0066] P543-1: Use the data verification set and data test set to test and verify the basic thermal power load prediction model to obtain model performance parameter information.

[0067] P543-2: Select a model optimizer based on the model performance parameter information.

[0068] P543-3: Use the model optimizer to optimize and update the model parameter information of the basic thermal power load prediction model to obtain the thermal power load adaptive prediction model.

[0069] Specifically, the input features of the data validation set and the data test set are respectively input into the basic thermal power load forecasting model to obtain the corresponding forecast output. By comparing the difference between the forecast output of the basic thermal power load forecasting model and the true value, the prediction error (such as mean square error), accuracy, recall rate and other evaluation indicators are calculated to obtain the model performance parameter information. These performance parameters reflect the performance of the model in different situations, such as the prediction accuracy under specific conditions.

[0070] According to the obtained model performance parameter information, select the most suitable optimizer to accelerate the model training process and improve the optimization effect. Different optimizers perform differently when processing different types of data. The Adam (adaptive momentum estimation) optimizer usually converges better when processing noisy data, while SGD (stochastic gradient descent) is suitable for processing simple data and has high computational efficiency. Exemplarily, if the mean square error of the model does not decrease significantly after multiple rounds of iterations, select an optimizer with an adaptive learning rate, such as the Adam optimizer. If the mean square error obtained by the model on the data validation set and the test set decreases rapidly in the early stage, but tends to stabilize and has a larger value in the later stage, and it is observed that the model oscillates during the training process, the Adadelta optimizer can be selected.

[0071] Use the selected optimizer to optimize the model, update the model parameters, and further improve the prediction performance. Through back propagation and gradient descent, the optimizer will gradually update the model parameters according to the model error until the loss function reaches the minimum value. For example, during the training process, if there is a large gap between the model prediction results and the actual thermal power load, the optimizer will adjust the network weights so that the next prediction is closer to the real data. Through the above steps, the thermal power load prediction model finally obtained has higher prediction accuracy and stronger adaptability, so that accurate thermal power load prediction results can be obtained in a variety of practical application environments.

[0072] Furthermore, the device described in the embodiment of the present application is also used to perform the following steps:

[0073] P5-1: Perform thermal power load prediction based on the thermal power load adaptive prediction model to obtain thermal power load prediction parameters.

[0074] P5-2: Perform system optimization analysis on the thermal power load prediction parameters to obtain a set of control optimization strategies for the cogeneration system.

[0075] P5-3: Based on the control optimization strategy set of the cogeneration system, the optimization effect prediction and strategy optimization of the cogeneration system are performed respectively to obtain the target cogeneration system control optimization strategy.

[0076] Specifically, the current influencing characteristic data, such as weather data and user behavior data, are obtained and input into the thermal power load adaptive prediction model. The model calculates and outputs the thermal power load prediction parameters based on the previously learned patterns and relationships. These thermal power load prediction parameters include the thermal power load amount (such as the predicted value of electricity and heat) in a certain period of time in the future, the change trend of the thermal power load (whether it is rising, falling or remaining stable), etc.

[0077] Determine the demand of the cogeneration system based on the thermal load forecast parameters. For example, if it is predicted that the electrical load will increase significantly while the thermal load is relatively stable, it is necessary to consider adjusting the power generation strategy of the cogeneration unit. Then, combined with the equipment characteristics, operating costs, energy supply conditions and other factors of the cogeneration system, a number of different cogeneration system control optimization strategies are formulated through mathematical models or algorithms (such as optimization algorithms, linear programming, etc.) to form a set of cogeneration system control optimization strategies. These strategies include how to adjust energy production, how to distribute loads, how to switch energy sources, etc. For example, if it is predicted that the load demand is high during a certain period of time, the proportion of solar energy and biomass energy generation can be adjusted to ensure efficient and uninterrupted energy supply.

[0078] For each cogeneration system control optimization strategy, use system simulation software or mathematical models to predict its optimization effect. For example, calculate indicators such as energy consumption, cost, and environmental emissions under each strategy. Then, based on the prediction results and pre-set optimization criteria (such as taking the lowest cost as the main criterion, while considering factors such as energy consumption and environmental emissions), these strategies are evaluated and optimized to select the most appropriate control strategy. The evaluation process can use multi-objective decision-making methods (such as hierarchical analysis method, fuzzy comprehensive evaluation method, etc.) to comprehensively evaluate the advantages and disadvantages of each strategy, so as to select the cogeneration system control strategy that best suits the current needs, ensure the reliability and economy of energy supply, and improve the overall performance of the system.

[0079] In summary, the thermal power load prediction device combined with demand-side energy analysis provided in the embodiments of the present application has the following technical effects:

[0080] The embodiment of the present application starts with data collection, and then goes through preprocessing, analysis, feature extraction, and finally constructs a model for load forecasting, forming a complete thermal power load forecasting system. By introducing demand-side energy analysis and influencing factor extraction, an adaptive thermal power load forecasting model is constructed from the perspective of demand-side energy consumption, which can more comprehensively capture the multi-dimensional load change rules, especially the impact of end-user behavior and external environmental changes on load demand, thereby improving the accuracy and reliability of thermal power load forecasting, better adapting to actual energy demand changes, and providing support for the stable operation of biomass and solar energy complementary cogeneration systems. At the same time, the adaptive deep learning model can optimize the prediction results in real time, adapt to load demand fluctuations in complex dynamic environments, and improve the flexibility and generalization ability of the prediction model, thereby achieving more scientific resource allocation in the operation and management of the cogeneration system, and optimizing energy utilization and load management.

[0081] Embodiment 2, as Figure 2 As shown, the embodiment of the present application provides a thermal power load prediction method combined with demand-side energy analysis, the method comprising:

[0082] Step S1: Collect and obtain a thermal power load data set of a biomass and solar energy complementary cogeneration system, wherein the thermal power load data set includes historical load data and energy consumption data on the terminal demand side.

[0083] Step S2: preprocessing the thermal and electric load data set to obtain a standard thermal and electric load data set.

[0084] Step S3: performing demand-side energy consumption analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set.

[0085] Step S4: extracting features from the standard thermal and electric load data set based on the thermal and electric load influencing factor set to obtain a thermal and electric load feature set.

[0086] Step S5: construct a deep learning model list, use the deep learning model list to train and optimize the thermal power load feature set, obtain a thermal power load adaptive prediction model, and perform thermal power load prediction based on the thermal power load adaptive prediction model.

[0087] Furthermore, step S2 of the embodiment of the present application further includes:

[0088] The thermal and electric load data set is normalized and time-aligned to obtain a thermal and electric load sequence data set; abnormal data is identified on the thermal and electric load sequence data set to obtain abnormal thermal and electric load data, wherein the abnormal thermal and electric load data includes inconsistent data, missing values, and out-of-range values; preprocessing and analyzing the abnormal thermal and electric load data to determine a data preprocessing step; and preprocessing the abnormal thermal and electric load data based on the data preprocessing step to obtain the standard thermal and electric load data set.

[0089] Furthermore, step S4 of the embodiment of the present application also includes:

[0090] Based on the thermal power load influencing factor set, feature extraction is performed on the standard thermal power load data set to obtain a load correlation influence feature set; the correlation degree of each influencing feature in the load correlation influence feature set is evaluated to obtain a load influence feature correlation degree set; according to the thermal power load forecast demand information, a feature correlation degree threshold is set; based on the feature correlation degree threshold, the load influence feature correlation degree set is screened to obtain the thermal power load feature set.

[0091] Furthermore, step S5 of the embodiment of the present application further includes:

[0092] Perform data characteristic analysis on the thermal power load feature set to obtain data characteristic information; determine data prediction demand characteristic information based on the thermal power load prediction demand information and the data characteristic information; perform matching and optimization based on the data prediction demand characteristic information and the deep learning model list to determine a target deep learning model; use the target deep learning model to train and optimize the thermal power load feature set to obtain the thermal power load adaptive prediction model.

[0093] Furthermore, the target deep learning model is used to train and optimize the thermal power load feature set to obtain the thermal power load adaptive prediction model, which also includes:

[0094] The thermal power load feature set is divided into data proportions to obtain a data training set, a data verification set and a data test set; the data training set is trained for feature recognition using the target deep learning model to generate a basic thermal power load prediction model; the basic thermal power load prediction model is verified and optimized using the data verification set and the data test set to obtain the thermal power load adaptive prediction model.

[0095] Furthermore, the basic thermal power load prediction model is verified and optimized by using the data verification set and the data test set to obtain the thermal power load adaptive prediction model, which also includes:

[0096] The basic thermal power load prediction model is tested and verified using the data verification set and the data test set to obtain model performance parameter information; a model optimizer is selected based on the model performance parameter information; the model optimizer is used to optimize and update the model parameter information of the basic thermal power load prediction model to obtain the thermal power load adaptive prediction model.

[0097] Furthermore, the method described in the embodiment of the present application also includes:

[0098] Based on the adaptive prediction model of thermal power load, thermal power load prediction is performed to obtain thermal power load prediction parameters; the thermal power load prediction parameters are systematically optimized to obtain a set of control optimization strategies for a combined heat and power system; based on the set of control optimization strategies for a combined heat and power system, optimization effect prediction and strategy optimization are performed on the combined heat and power system to obtain a target combined heat and power system control optimization strategy.

[0099] Through the above-mentioned detailed description of the thermal power load prediction device combined with demand-side energy analysis in this specification, those skilled in the art can clearly understand the thermal power load prediction method combined with demand-side energy analysis in this embodiment. For the method disclosed in Example 2, since it corresponds to the device disclosed in Example 1 and has corresponding execution steps and beneficial effects, the relevant parts can be referred to the description of the device part.

[0100] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A thermal power load forecasting device combined with demand-side energy analysis, characterized in that: The device comprises: A data acquisition module, used to acquire a thermal load data set of a biomass and solar energy complementary cogeneration system, wherein the thermal load data set includes historical load data and energy consumption data on the terminal demand side; A data preprocessing module, used to preprocess the thermal and electric load data set to obtain a standard thermal and electric load data set; A data analysis module, used to perform demand-side energy analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set; A data feature extraction module, used for extracting features from the standard thermal load data set based on the thermal load influencing factor set to obtain a thermal load feature set; The load prediction module is used to construct a deep learning model list, use the deep learning model list to train and optimize the thermal power load feature set, obtain a thermal power load adaptive prediction model, and perform thermal power load prediction based on the thermal power load adaptive prediction model.

2. The thermal power load prediction device combined with demand-side energy analysis according to claim 1, characterized in that: The data preprocessing module is also used to perform the following steps: Normalizing and time-aligning the thermal and electric load data set to obtain a thermal and electric load sequence data set; Performing abnormal data identification on the thermal load sequence data set to obtain abnormal thermal load data, wherein the abnormal thermal load data includes inconsistent data, missing values, and out-of-range values; Performing preprocessing analysis on the abnormal thermal and electric load data to determine data preprocessing steps; The abnormal thermal and electric load data is preprocessed based on the data preprocessing step to obtain the standard thermal and electric load data set.

3. The thermal power load prediction device combined with demand-side energy analysis according to claim 1, characterized in that: The data feature extraction module is also used to perform the following steps: Based on the thermal power load influencing factor set, feature extraction is performed on the standard thermal power load data set to obtain a load correlation influence feature set; Evaluate the correlation degree of each influencing feature in the load correlation influencing feature set to obtain a load influencing feature correlation degree set; According to the forecast demand information of thermal power load, the characteristic correlation threshold is set; The load impact feature correlation set is screened based on the feature correlation threshold to obtain the thermal power load feature set.

4. The thermal power load prediction device combined with demand-side energy analysis according to claim 3, characterized in that: The load forecasting module is also used to perform the following steps: Performing data characteristic analysis on the thermal power load characteristic set to obtain data characteristic information; Determining data prediction demand characteristic information according to the thermal power load prediction demand information and the data characteristic information; Based on the data prediction demand characteristic information and the deep learning model list, a matching optimization is performed to determine a target deep learning model; The target deep learning model is used to train and optimize the thermal power load feature set to obtain the thermal power load adaptive prediction model.

5. The thermal power load prediction device combined with demand-side energy analysis according to claim 4, characterized in that: The load forecasting module is also used to perform the following steps: Dividing the thermal power load feature set into data proportions to obtain a data training set, a data verification set, and a data test set; Using the target deep learning model to perform feature recognition training on the data training set to generate a basic thermal power load prediction model; The data verification set and the data test set are used to verify and optimize the basic thermal power load prediction model to obtain the thermal power load adaptive prediction model.

6. The thermal power load prediction device combined with demand-side energy analysis according to claim 5, characterized in that: The load forecasting module is also used to perform the following steps: The basic thermal power load prediction model is tested and verified using the data verification set and the data test set to obtain model performance parameter information; Selecting a model optimizer according to the model performance parameter information; The model optimizer is used to optimize and update the model parameter information of the basic thermal power load prediction model to obtain the thermal power load adaptive prediction model.

7. The thermal power load prediction device combined with demand-side energy analysis according to claim 1, characterized in that: The device is also used to perform the following steps: Performing thermal power load prediction based on the thermal power load adaptive prediction model to obtain thermal power load prediction parameters; Performing system optimization analysis on the thermal power load prediction parameters to obtain a control optimization strategy set for the combined heat and power system; Based on the control optimization strategy set of the cogeneration system, optimization effect prediction and strategy optimization are performed on the cogeneration system respectively to obtain the target cogeneration system control optimization strategy.

8. A thermal power load forecasting method combined with demand-side energy analysis is characterized in that: The method is performed by the device according to any one of claims 1 to 7, comprising: Collect and obtain a thermal power load data set of a biomass and solar energy complementary cogeneration system, wherein the thermal power load data set includes historical load data and energy consumption data on the terminal demand side; Preprocessing the thermal and electric load data set to obtain a standard thermal and electric load data set; Performing demand-side energy consumption analysis and influencing factor extraction on the standard thermal power load data set to obtain a thermal power load influencing factor set; Extracting features from the standard thermal and electric load data set based on the thermal and electric load influencing factor set to obtain a thermal and electric load feature set; A deep learning model list is constructed, the deep learning model list is used to train and optimize the thermal power load feature set, a thermal power load adaptive prediction model is obtained, and thermal power load prediction is performed based on the thermal power load adaptive prediction model.

Images