Digital Park Carbon Emission Trend Deduction Method, System and Electronic Device

Abstract

The present invention discloses a method, system and electronic device for deducing the carbon emission trend of a digital park, which relates to the technical field of dynamic optimization of carbon emissions. The method includes: by combining the production plan and climate characteristics of an industrial park in a future time period, predicting the energy consumption sequence at P consecutive time points, and simultaneously predicting the output power sequence of multi-source energy based on historical power generation records; on this basis, analyzing the energy supply ratio and generating a power supply ratio sequence, and finally integrating the energy consumption sequence and the power supply ratio sequence by using the carbon emission factor per unit of energy to calculate the carbon emissions in the future time zone, so as to achieve accurate deduction of the carbon emission trend. It solves the technical problems of insufficient collaborative modeling of multi-source energy and lagging dynamic optimization of power supply ratio in the existing carbon emission prediction of parks, and achieves the technical effects of reducing the carbon emission prediction error rate and improving the dynamic optimization response speed of the new energy power supply ratio.

Description

Technical Field

[0001] This application relates to the technical field of dynamic optimization of carbon emissions, and particularly to a method, system and electronic device for deducing the carbon emission trend in a digital park. Background Art

[0002] The existing carbon emission management in parks generally relies on static historical data and a single energy model, making it difficult to cope with the complex challenges brought about by dynamic adjustments of production plans, climate fluctuations, and coordinated supply of multiple energy sources. Due to the lack of refined modeling of the linkage between production intensity, environmental temperature, and energy supply and demand, traditional prediction methods result in low accuracy of carbon emission prediction and are unable to respond to changes in the energy structure in real time. How to achieve high-precision carbon emission deduction and dynamic energy scheduling under the coupling of multiple factors has become a technical problem to be solved urgently.

[0003] In the related technologies at the present stage, there are technical problems of insufficient collaborative modeling of multiple energy sources and lagging optimization of dynamic power supply ratios in the prediction of carbon emissions in parks. Summary of the Invention

[0004] This application solves the technical problems of insufficient collaborative modeling of multiple energy sources and lagging optimization of dynamic power supply ratios in the existing carbon emission prediction in parks by providing a method, system and electronic device for deducing the carbon emission trend in a digital park.

[0005] This application provides a method for deducing the carbon emission trend in a digital park, including:

[0006] According to the production plan and climate characteristics of the industrial park in a predetermined future time zone, predict the energy consumption at P consecutive time points to generate a predicted energy consumption sequence; according to the historical power generation records, predict the multi-energy output power at P consecutive time points to generate multiple predicted energy output power sequences; based on the predicted energy consumption sequence and multiple predicted energy output power sequences, analyze the energy supply ratio to generate a predicted energy supply ratio sequence; calculate the carbon emissions in the predetermined future time zone according to the unit energy carbon emission factor, the predicted energy consumption sequence and the predicted energy supply ratio sequence, and obtain the predicted carbon emissions as the result of carbon emission trend deduction.

[0007] This application provides a system for deducing the carbon emission trend in a digital park, including:

[0008] A predicted energy consumption sequence generation module, which is used to predict the energy consumption at consecutive P time points according to the production plan and climate characteristics of the industrial park in a predetermined future time zone, and generate a predicted energy consumption sequence; a predicted energy output power sequence generation module, which is used to predict the multi-energy output power at consecutive P time points according to the historical power generation records, and generate multiple predicted energy output power sequences; an energy supply ratio analysis module, which is used to analyze the energy supply ratio based on the predicted energy consumption sequence and multiple predicted energy output power sequences, and generate a predicted energy supply ratio sequence; a carbon emission calculation module, which is used to calculate the carbon emissions in the predetermined future time zone according to the unit energy carbon emission factor, the predicted energy consumption sequence and the predicted energy supply ratio sequence, and obtain the predicted carbon emissions as the carbon emission trend deduction result.

[0009] The present application also provides an electronic device, including:

[0010] A memory for storing executable instructions; a processor for implementing the carbon emission trend deduction method for a digital industrial park when executing the executable instructions stored in the memory.

[0011] It is intended to use the carbon emission trend deduction method, system and electronic device proposed in the present application. First, by combining the production plan and climate characteristics of the industrial park in the future period, the energy consumption sequence at consecutive P time points is predicted, and at the same time, the output power sequence of multi-source energy is predicted based on the historical power generation records; on this basis, the energy supply ratio is analyzed and the power supply ratio sequence is generated. Finally, the unit energy carbon emission factor is used to integrate the energy consumption sequence and the power supply ratio sequence, and the carbon emissions in the future time zone are calculated to achieve accurate deduction of the carbon emission trend, and the technical effects of reducing the carbon emission prediction error rate and improving the dynamic optimization response speed of the new energy power supply ratio are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in the present application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the operations above or below do not necessarily need to be executed precisely in sequence. On the contrary, as needed, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0013] Figure 1 It is a schematic flowchart of the carbon emission trend deduction method for a digital industrial park provided by an embodiment of the present application;

[0014] Figure 2Schematic diagram of the digital park carbon emission trend deduction system provided by the embodiments of the present application;

[0015] Figure 3 Schematic diagram of the exemplary electronic device of the present application.

[0016] Explanation of reference numerals: Prediction energy consumption sequence generation module 10, prediction energy output power sequence generation module 20, energy power supply ratio analysis module 30, carbon emission calculation module 40, input device 201, processor 202, memory 203, output device 204. Detailed implementation manners

[0017] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically gives the detailed implementation manners of the present application.

[0018] In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0019] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first\second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0020] The embodiments of the present application provide a digital park carbon emission trend deduction method, as Figure 1 shown, the method includes:

[0021] Step S100: According to the production plan and climate characteristics of the industrial park in a predetermined future time zone, conduct energy consumption prediction at consecutive P time points to generate a predicted energy consumption sequence. Specifically, first obtain the production plan in the predetermined future time zone from the industrial park management department, such as the expected product output, equipment operation arrangements, etc. At the same time, master the climate characteristic data with the help of the meteorological department or professional models. Set consecutive P (P>30) time points at specific intervals in this time zone and divide them into P time intervals. For each interval, considering factors such as the energy consumption difference in product production processes and the impact of climate on equipment operation, combine and divide the production plan and climate characteristics to determine P energy consumption correlation data. Construct a sample energy consumption correlation data set and a sample energy consumption amount set by collecting past production records. Configure K (K≥3) prediction operators using machine learning, conduct supervised training and verification on them. After the loss function converges, integrate and construct an energy consumption prediction plug-in. Use this plug-in to predict the energy consumption amount according to the P energy consumption correlation data respectively, convert it into the power mean, and arrange it in chronological order to generate a predicted energy consumption sequence, laying the energy consumption data foundation for the deduction of carbon emission trends.

[0022] In a possible implementation manner, according to the production plan and climate characteristics of the industrial park in a predetermined future time zone, conduct energy consumption prediction at consecutive P time points to generate a predicted energy consumption sequence. Step S100 further includes step S110: Obtain the production plan and climate characteristics of the industrial park in a predetermined future time zone, where the production plan is the expected product output and the climate characteristic is the expected environmental temperature. Specifically, to obtain the production plan (expected product output) and climate characteristics (expected environmental temperature) of the industrial park in a predetermined future time zone, multiple approaches are needed. In terms of obtaining the production plan, on the one hand, connect with the production management departments of each enterprise in the park and the park management agency, collect information such as the expected output, time distribution, and production capacity adjustment plan of each product in the future predetermined time period, and then cross-compare and verify the data, and judge the rationality in combination with past production data; in terms of obtaining climate characteristics, cooperate with the meteorological bureau or meteorological service company to sign a data service agreement to obtain refined temperature predictions, and at the same time use the data of the park's own meteorological monitoring station. When necessary, temporarily deploy simple equipment for assistance. Finally, integrate the two types of data, use tools for statistical analysis and correlate with the production plan to lay a solid data foundation for subsequent energy consumption and other analyses.

[0023] Step S120, set consecutive P time points within the predetermined future time zone at a predetermined time interval, and divide to obtain P time intervals, where P is an integer greater than 30. Specifically, determining the time interval requires a comprehensive analysis of the purpose, data characteristics, and acquisition and processing capabilities. For scenarios such as chemical production with large fluctuations in energy consumption, a minute-level interval may be selected, while for stable production activities such as textile production, an hour-level interval is sufficient. After determining the interval, within the predetermined future time zone, starting from the starting moment, set consecutive P (P > 30) time points at the interval. For example, set 168 points at 1-hour intervals for the next week. Based on these time points, time intervals are formed between adjacent points. In the above example, 168 intervals are divided. Each interval can determine the energy consumption-related data in combination with production plans, climate conditions, etc. Since P is large enough, it can comprehensively reflect the changes of various factors at different times and improve the reliability of the analysis results.

[0024] Step S130, divide and combine the production plan and climate characteristics according to the P time intervals to determine P energy consumption-related data. Specifically, after determining the P time intervals, divide and combine the production plan and climate characteristics to determine the energy consumption-related data. Regarding the production plan, break down the expected product output by time interval. For example, in an electronics manufacturing park, the monthly output is allocated by day, and at the same time, the energy consumption characteristics of each production link are associated, considering changes in the operating state of equipment, such as the impact of equipment maintenance shutdown on energy consumption. In terms of climate characteristics, obtain the ambient temperature by time interval. For example, calculate the daily average, maximum, and minimum temperatures from hourly temperature data, analyze the impact of temperature on equipment and production, and combine seasonal factors. Finally, establish a calculation model for energy consumption-related data that comprehensively considers production plan and climate characteristic factors, and calculate data such as the total energy consumption and energy consumption per unit product corresponding to each time interval according to the model, so as to determine P energy consumption-related data and lay a solid data foundation for subsequent energy consumption prediction and analysis.

[0025] Step S140, using the energy consumption prediction plug-in, according to the P energy consumption related data, respectively predict, output P predicted energy consumption, calculate P predicted energy consumption power means, and construct the predicted energy consumption sequence. Specifically, when using the energy consumption prediction plug-in to construct the predicted energy consumption sequence, first comprehensively check and debug the plug-in trained based on a large amount of historical energy consumption and related influencing factors data. After that, the energy consumption related data of P comprehensive production plans and climate characteristics and other factors are input into the plug-in in turn, and the algorithm model in the plug-in parses the data, and weights each factor according to the weight distribution method obtained by training, and outputs P predicted energy consumption after complex calculations. Then, use the formula "predicted energy consumption power mean = predicted energy consumption ÷ corresponding time interval duration" to calculate P predicted energy consumption power means. Finally, arrange these means according to the order of time intervals, and construct a predicted energy consumption sequence that can intuitively show the predicted trend of energy consumption power changes in the predetermined future time zone, providing strong support for park energy planning and other work.

[0026] In a possible implementation, an energy consumption prediction plug-in is used to predict the P energy consumption related data, output P predicted energy consumptions, calculate the P predicted energy consumption power means, and construct the predicted energy consumption sequence. Step S140 further includes step S141, based on the historical production records of the industrial park, a sample energy consumption related data set is collected, and the energy consumption corresponding to different energy consumption related data is counted according to the historical electricity consumption records, which is set as the sample energy consumption, and a sample energy consumption set is constructed. Specifically, collaborate with the data management and power management departments of the industrial park to clarify the database location and access rights of the historical production records, and obtain historical electricity consumption records from the power management system or supplier. Samples are extracted from the historical production records at certain time intervals, and data such as product production, equipment operation, and production process are collected. Factors affecting energy consumption such as the production workshop environment and logistics transportation are considered to integrate and form a sample energy consumption related data set. Then, using time as the matching dimension, the associated data is matched with the historical electricity consumption records, and the energy consumption corresponding to each set of associated data is counted according to the power circuit identification of the production device, which is set as the sample energy consumption. After checking and correcting the abnormal data, a reliable sample energy consumption set is constructed according to the order of the associated data to provide high-quality data support for subsequent model training.

[0027] Step S142: Use the sample energy consumption correlation dataset and the sample energy consumption volume dataset as training data to train multiple prediction operators respectively, and obtain multiple energy consumption prediction branches. Specifically, when constructing an energy consumption prediction model, first screen diverse prediction operators such as linear regression, decision tree, and neural network, determine K (K≥3) of them and initialize them respectively. For example, initialize the coefficient vector of linear regression, set parameters such as the maximum depth of the decision tree, and determine the structure of the neural network and initialize the weights and biases. Then preprocess the sample energy consumption correlation dataset and the sample energy consumption volume dataset, standardize the numerical data, handle the missing values, and then divide them into a training set and a validation set according to the ratio of 70%-80% and 20%-30%. Then start a training loop for each prediction operator. The neural network calculates the loss between the predicted value and the true value through forward propagation and updates the parameters through backward propagation; linear regression minimizes the loss function to solve the coefficients; the decision tree selects features and split points to build a tree. During training, regularly evaluate with the validation set, and adjust in time when problems such as overfitting occur. For example, reduce the learning rate of the neural network, improve the model of linear regression, and prune the decision tree. When the performance of the prediction operator reaches the standard in the validation set and the training converges, save the trained operator to form an energy consumption prediction branch. Finally, obtain K different prediction branches for subsequent plug-in integration.

[0028] Step S143: Integrate and construct the energy consumption prediction plug-in according to the multiple energy consumption prediction branches, where the output of the energy consumption prediction plug-in is the mean value of the outputs of the multiple energy consumption prediction branches. Specifically, to construct an energy consumption prediction plug-in to integrate multiple energy consumption prediction branches, first, it is necessary to clarify the characteristics of each branch. For example, the linear regression branch is good at processing linear trend data, the decision tree branch can handle discrete data and complex decision logics, and the neural network branch has strong non-linear fitting ability. Then design the integration architecture, build the main program framework with modular programming, which is responsible for the adaptation, distribution of input data, and collection of the prediction results of each branch. After the new energy consumption correlation data is input into the plug-in, the main program converts the data according to the requirements of each branch and distributes it in parallel using multi-threading or distributed technology. Each branch receives the data and completes the operation. Branches such as linear regression, decision tree, and neural network output predicted values according to their respective model algorithms. The main program collects the results and calculates the arithmetic mean according to the setting. For example, if the outputs of the three branches are 100, 120, and 110 kWh respectively, the final output is (100 + 120 + 110) ÷ 3 = 110 kWh for subsequent work such as energy planning.

[0029] In a possible implementation, the sample energy consumption correlation dataset and the sample energy consumption quantity dataset are used as training data to train multiple prediction operators respectively, and multiple energy consumption prediction branches are obtained. Step S142 further includes step S1421, where K prediction operators are configured based on machine learning, where K is greater than or equal to 3. Specifically, when configuring K (K≥3) prediction operators based on machine learning, first clarify the business requirements of energy consumption prediction, such as prediction duration, object scope, etc., and deeply analyze the data characteristics of the sample energy consumption correlation dataset and the sample energy consumption quantity dataset, covering scale, type, distribution, and feature correlation. According to the analysis results, select appropriate operator types. Linear regression is suitable for linear relationship scenarios, decision trees can handle complex data and non-linear relationships, neural networks are good at capturing complex patterns, support vector machines perform well in small sample high-dimensional data, and random forests can improve stability. After determining the type, set key parameters for each operator, such as the regularization parameter of linear regression, the depth, sample number, and splitting criterion of decision trees, the network structure, activation function and other parameters of neural networks, the kernel function and other parameters of support vector machines, and the number of decision trees and other parameters of random forests. For complex models such as neural networks and random forests, design reasonable architectures. For the former, determine the number of neurons in the input, hidden, and output layers, etc., and for the latter, control the decision tree generation process. Finally, use indicators such as mean squared error to evaluate the performance of the operators through cross-validation, and adjust and optimize the parameters using methods such as grid search according to the results, laying a solid foundation for the training and integration of the energy consumption prediction model.

[0030] Step S1422, divide the training data into K equal parts, and perform supervised training and verification on the K prediction operators respectively until the loss function converges, obtaining multiple energy consumption prediction branches. Specifically, in the process of constructing the energy consumption prediction branches, first ensure that the training data is accurate and complete. Use algorithms such as stratified random sampling to divide the sample energy consumption correlation dataset and the sample energy consumption quantity dataset into K equal parts to ensure that the data feature distributions of each part are similar. Subsequently, for each part of the data, perform supervised training on the K prediction operators respectively. Linear regression adjusts the coefficients by the least squares method, decision trees are constructed according to criteria such as information gain, and neural networks update the parameters by the backpropagation algorithm. The mini-batch gradient descent method is often used during training. During training, reserve 10%-20% of the training data as the validation set, and evaluate the performance using indicators such as mean absolute error. Linear regression can adjust the regularization parameter, decision trees can be pruned and optimized, and neural networks can change the learning rate, etc. Continuously train and verify until the loss functions of the K prediction operators all converge. At this time, each operator forms an energy consumption prediction branch. They are based on different algorithms and training data, and will cooperate to improve the prediction accuracy in the subsequent integration of the energy consumption prediction plug-in.

[0031] Step S200: Based on historical power generation records, perform multi - energy output power prediction at consecutive P time points to generate multiple predicted energy output power sequences. Specifically, when performing multi - energy output power prediction and generating prediction sequences, first, interface with relevant institutions to obtain historical power generation records covering various energy - generating facilities, comprehensively check data integrity, unify the format, and pre - process the data, such as repairing missing values, handling outliers, and normalizing the data. Then, set the prediction time range according to actual needs. For example, to arrange power dispatching, predict the next week. Then, select the time - point interval according to the energy fluctuation characteristics and accuracy requirements. If the interval is one hour and predicting for a week, then P is 168. After that, select models according to different energy characteristics. For example, use the Holt - Winters model for thermal power generation, neural networks for wind power and photovoltaic power, and a combined physical and data - based method for hydropower. Model integration optimization can also be carried out. After preparing 70% - 80% of the training set and 20% - 30% of the validation set, train the model, continuously adjust the parameters, and use the validation set to prevent overfitting. After the model is trained, input future data for prediction. Finally, organize the prediction results, store them according to energy and time, generate and visualize the predicted output power sequences for each energy, providing an intuitive reference for decision - making.

[0032] In a possible implementation, based on historical power generation records, perform multi - energy output power prediction at consecutive P time points to generate multiple predicted energy output power sequences. Step S200 further includes step S210: Obtain the multi - source power supply mode of the industrial park and multiple historical power generation records of the multi - source power supply mode. Specifically, to obtain the multi - source power supply mode of the industrial park and its historical power generation records, first, clarify the power supply types such as photovoltaic, wind power, energy storage, etc., and the time range for collecting data. If it is for short - term dispatching, collect high - frequency data from the recent few months to one year. If it is for long - term planning, collect data for several years. Then, build a data acquisition channel. Internally, interface with the operation and maintenance management teams of photovoltaic power plants, wind farms, and energy storage facilities, and extract information such as power generation power, charge - discharge data, and equipment parameters at different time periods from their monitoring systems or energy management systems. Externally, obtain meteorological data such as sunlight and wind speed from the meteorological department and power market data from the power trading platform. Finally, unify the data format, standardize the power generation power unit, time format, etc., verify and clean the data, fill in missing values, correct outliers, and compare with the actual equipment operation and physical laws to ensure the accuracy and reliability of the data.

[0033] Step S220: Obtain P power generation impact data sets of the multi-source power supply method within P time intervals. Specifically, to obtain P power generation impact data sets of the multi-source power supply method within P time intervals, the time intervals need to be determined first according to the analysis purpose, business, and data characteristics. For short-term scheduling, the unit is hours or minutes, and for long-term planning, the unit is days, weeks, or months, and it is set in combination with the park's production operation and data acquisition and storage capabilities. Then, identify the influencing factors. Photovoltaic power generation is affected by light intensity, ambient temperature, and air quality. Wind power depends on wind speed, wind direction, and air density. Energy storage is related to charge and discharge strategies, remaining battery capacity, and equipment aging degree. Then, establish a data acquisition system. Meteorological factors are obtained from meteorological monitoring stations, energy storage data is obtained from the energy management system, and air quality data is collected in cooperation with the environmental protection department or using professional equipment. Set the acquisition frequency according to the time interval and ensure the time synchronization of the equipment. Finally, clean and preprocess the data, remove outliers, handle missing values, and standardize it. Then, divide it according to the time interval to generate P power generation impact data sets containing various influencing factor data of the multi-source power supply method, providing input for the subsequent construction of the integrated power predictor.

[0034] Step S230: Based on historical power generation records, construct an integrated power predictor using machine learning, and perform multi-energy output power prediction respectively according to the P power generation impact data sets to generate multiple predicted energy output power sequences. Specifically, first collect historical power generation records of multi-source power supply such as photovoltaic, wind power, and energy storage from multiple channels to ensure integrity and accuracy. Clean the data, fill in missing values, correct outliers, and extract relevant features such as light duration. After encoding and conversion, perform standardization or normalization. Then, construct an integrated power predictor using machine learning. Select algorithms such as RNN, LSTM, SVM, and random forest regression, determine the model structure parameters, optimize them through methods such as grid search, divide the training set, validation set, and test set, train the model, and adjust it according to the validation set to prevent overfitting. Evaluate the generalization ability using the test set. Finally, preprocess the P power generation impact data sets to make them consistent with the training set and then input them into the model. The model predicts the multi-energy output power of each time interval according to the learned patterns and generates multiple predicted energy output power sequences of photovoltaic, wind power, energy storage, etc. in time order, providing a decision-making basis for park energy planning, etc.

[0035] Step S300: Based on the predicted energy consumption sequence and multiple predicted energy output power sequences, perform an analysis of the energy supply ratio, and generate a predicted energy supply ratio sequence. Specifically, when performing the analysis of the energy supply ratio and generating the predicted energy supply ratio sequence based on the predicted energy consumption sequence and multiple predicted energy output power sequences, it is first necessary to carefully verify the data, correct outliers, ensure that the timestamps match, obtain the unit energy carbon emission factor from authoritative channels and unify the dimension. Then, calculate the power supply amount of each power supply method at each time point according to the predicted energy output power and the time interval, sum them to obtain the total power supply amount, and further calculate the power supply ratio of each power supply method. Next, combine the power supply ratio with the unit energy carbon emission factor to analyze the proportion of the carbon emission contribution of each power supply method. Finally, arrange the power supply ratios at each time point in time order to generate a predicted energy supply ratio sequence, which can intuitively present the changes in the power supply ratios of different power supply methods at different times, provide a reference for energy planners, and help optimize the energy supply strategy.

[0036] In a possible implementation manner, based on the predicted energy consumption sequence and multiple predicted energy output power sequences, perform an analysis of the energy supply ratio, and generate a predicted energy supply ratio sequence. Step S300 further includes step S310: Select the average value of the first predicted energy consumption power and the first set of predicted energy output powers at the first time point. Specifically, when selecting the first time point, it is necessary to be based on the analysis purpose. For example, to optimize the energy dispatch during the peak electricity consumption period, 10 am on a weekday can be selected; to focus on the utilization of new energy at night, 3 am can be selected; it may also be affected by specific production processes such as factory equipment maintenance. After determining the time point, extract data from the predicted energy consumption sequence. If there are multiple predicted values, use the arithmetic mean method to calculate the average value of the first predicted energy consumption power. For example, the average value is calculated from 5 predicted values. Then, for the predicted energy output power sequences of multi-source power supply methods such as photovoltaic, wind power, and energy storage, extract the predicted values at the first time point respectively. For example, the photovoltaic value is obtained according to the light model, the wind power value is based on the wind speed model, and the energy storage value is obtained from the charge and discharge sequence. Integrate these values into the first set of predicted energy output powers to provide key data for the subsequent analysis of the energy supply ratio.

[0037] Step S320: Configure the first electricity demand index based on the first product output and the first environmental temperature in the first time interval. Specifically, when configuring the first electricity demand index, first define the first time interval according to the energy analysis time scale and objectives, extract the first product output data in this interval from the production record database, and at the same time obtain the first environmental temperature data from the meteorological monitoring equipment or relevant platforms. The multi-monitoring point data is processed by arithmetic or weighted average method according to requirements. Then analyze the relationship between product output and electricity demand, establish a correlation model using methods such as linear regression, and consider factors such as production process to correct the model; then study the relationship between environmental temperature and electricity demand, establish a temperature-electricity demand function, pay attention to the impact of temperature fluctuations and determine the fluctuation impact coefficient. Finally, determine the weights according to the importance of the impact of product output and environmental temperature on electricity demand, calculate the first electricity demand index by weighted summation, compare it with the actual electricity consumption data for verification. If the deviation is large, check the data, model and weight settings, and adjust until the index can accurately reflect the electricity demand in this interval.

[0038] Step S330: Taking the first electricity demand index as a constraint and aiming at the maximum proportion of new energy power generation, conduct an energy power supply ratio analysis according to the first predicted energy consumption power mean and the first predicted energy output power set, output the first predicted energy power supply ratio, and add it to the predicted energy power supply ratio sequence. Specifically, to achieve the goal of the maximum proportion of new energy power generation while meeting the constraint of the first electricity demand index, an optimization model needs to be constructed. Set the proportion of new energy power supply such as photovoltaic, wind power, and energy storage as decision variables, with values between 0 and 1 and the sum being 1. The objective function is to maximize the sum of the proportion of new energy power supply. Take the first electricity demand index, the output power of each power supply method and its own physical limitations as constraint conditions. Select a linear programming algorithm such as the simplex method, transform the model into a standard form and solve it. By iteratively adjusting the variable values until the optimality conditions are met, a set of new energy power supply ratios such as 30% for photovoltaic and 40% for wind power are obtained. Organize this result into the first predicted energy power supply ratio for output, and add it to the predicted energy power supply ratio sequence in chronological order to provide dynamic data support for energy planning.

[0039] In a possible implementation, the first electricity demand index is configured according to the first product output and the first environmental temperature in the first time interval. Step S320 further includes step S321, calculating the ratio of the first product output to the average product output in the historical time interval, which is set as the first production intensity. Specifically, to calculate the first production intensity, the historical time interval needs to be determined first according to the analysis purpose and data characteristics. If the impact of short-term production fluctuations on energy demand is concerned, a similar recent period can be selected, such as the same hourly period of each day in the past week as the first time interval; if a long-term trend analysis is carried out, a wider cycle should be covered, such as the same week in the same month of the past quarter. When selecting the interval, it is necessary to ensure that the data is available and stable, and avoid interference periods such as equipment maintenance. Then, collect the product output data in the historical time interval from the production management system, etc., unify the format and clean it, and select methods such as arithmetic mean, weighted mean or median according to the data distribution to calculate the average product output. Finally, obtain the product output in the first time interval, divide it by the average value to get the first production intensity, which can reflect the change of the production intensity in the first time interval relative to the historical average level.

[0040] Step S322, calculating the ratio of the first environmental temperature to the average environmental temperature in the historical time interval, which is set as the first environmental intensity. Specifically, when calculating the first environmental intensity, first determine the historical time interval according to the analysis target. If short-term impacts are concerned, select a similar recent period; if long-term trends are studied, cover a long cycle. At the same time, avoid periods of abnormal temperature fluctuations to ensure stable and reliable data acquisition. Then, collect data from multiple channels such as official meteorological monitoring stations, the park's own equipment and professional data service providers, and check and clean outliers and missing values. After that, select methods such as arithmetic mean or weighted mean according to the data characteristics to calculate the average value, and verify and adjust it with historical climate data. Finally, obtain the first environmental temperature from the same reliable source, divide it by the average value to get the first environmental intensity, which reflects the deviation degree of the environmental temperature in the first time interval relative to the historical average level.

[0041] Step S323, calculating the first electricity intensity according to the first production intensity and the first environmental intensity, and setting the first power supply reliability coefficient according to the first electricity intensity as the first electricity demand index, where the power supply reliability coefficient is positively correlated with the electricity intensity. Specifically, to calculate the first electricity intensity, first determine the weights of the first production intensity and the first environmental intensity through the correlation analysis of historical electricity consumption, production and environmental temperature data, combined with expert experience, such as setting them as 0.6 and 0.4 respectively. Use the weighted summation model, that is

[0042] ,

[0043] Substitute the obtained intensity values to calculate the first electricity intensity, and get 1.122. Since the power supply reliability coefficient is positively correlated with the electricity intensity, construct a linear function , the values of k and b are determined by regression analysis of historical data. For example, k = 0.8 and b = 0.1. Substitute the first electricity consumption intensity into the function to calculate the first power supply reliability coefficient. For example, 0.9976 is obtained, which is used as the first electricity demand index to provide a key basis for energy decision-making.

[0044] In a possible implementation, with the first electricity demand index as a constraint and the maximum proportion of new energy power generation as the goal, based on the first predicted energy consumption power mean and the first predicted energy output power set, perform an analysis of the energy supply proportion, and output the first predicted energy supply proportion, which is added to the predicted energy supply proportion sequence. Step S330 further includes step S331. According to the historical power generation records, respectively count the stable power supply proportions of each power supply method in the multi-source power supply method within a predetermined historical time period, which are set as historical power supply reliability coefficients, and obtain multiple historical power supply reliability coefficients. Specifically, to calculate the historical power supply reliability coefficient, first collect the power generation records accurate to each time point of each multi-source power supply method (such as photovoltaic, wind power, energy storage) within a predetermined historical time period from multiple channels such as the energy management system and the power generation equipment log, and clean and correct errors, anomalies, and missing values. Then define the stable power supply state for different power supply methods. For example, the photovoltaic power generation power fluctuates within ±10% of the rated power, the wind power is within the normal operating power range, and the energy storage charge and discharge power is within the safe range and can stably output. Subsequently, traverse the data point by point according to this standard, accumulate the stable power supply duration. For example, the photovoltaic is 6000 hours, the wind power is 5000 hours, and the energy storage is 4000 hours in the past year. Finally, divide the stable power supply duration of each power supply method by the total duration to obtain the historical power supply reliability coefficient. For example, the photovoltaic is about 0.685, the wind power is about 0.571, and the energy storage is about 0.457. The coefficients are subsequently used for the analysis of the energy supply plan.

[0045] Step S332, with the goal of meeting the first predicted energy consumption power mean, randomly generate the first initial energy supply plan according to the predicted energy output power set, where the initial energy supply plan includes the energy supply proportion. Specifically, when generating the first initial energy supply plan, with the goal of meeting the first predicted energy consumption power mean, rely on the set of predicted output powers of multi-source power supply methods including photovoltaic, wind power, energy storage, etc. to carry out the work. Through the random number generation algorithm, randomly allocate the energy supply proportion for each power supply method within the range of 0 to 1, and ensure that the sum of all proportions is 1. During this process, generate and verify the rationality of the proportion multiple times. According to the generated proportion and the predicted output power, calculate the power supply power of each power supply method. For example, the photovoltaic power supply power is the product of its predicted output power and the proportion. Then add up the power supply powers of each power supply method to obtain the total power supply power, and compare it with the first predicted energy consumption power mean. If it does not meet the requirement, adjust the proportion and recalculate until the total power supply power meets the target requirements.

[0046] Step S333: Using the energy supply ratio as the weight, calculate the first overall power supply reliability coefficient of the first initial energy supply plan based on the multiple historical power supply reliability coefficients. Specifically, to calculate the first overall power supply reliability coefficient, first determine the energy supply ratio corresponding to each power supply method from the first initial energy supply plan. For example, in a plan with three power supply methods: photovoltaic, wind power, and energy storage, their respective energy supply ratios are 0.3, 0.4, and 0.3. At the same time, we need to review the historical power supply reliability coefficients of each power supply method obtained from historical power generation records. For example, the historical power supply reliability coefficient of photovoltaic is 0.685, that of wind power is 0.75, and that of energy storage is 0.8. Next, according to the weighted calculation method, multiply the energy supply ratio of each power supply method by its corresponding historical power supply reliability coefficient. Taking photovoltaic as an example, multiply the energy supply ratio of photovoltaic, 0.3, by its historical power supply reliability coefficient, 0.685, to get 0.2055. Similarly, for wind power, it is 0.4 multiplied by 0.75, and the result is 0.3; for energy storage, it is 0.3 multiplied by 0.8, which equals 0.24. Finally, add up these product results, that is, 0.2055 plus 0.3 plus 0.24, and the sum obtained, 0.7455, is the first overall power supply reliability coefficient. This coefficient reflects the overall power supply reliability level of the first initial energy supply plan considering the power supply ratios of each power supply method and their historical power supply reliability levels.

[0047] Step S334: If the first overall power supply reliability coefficient is greater than or equal to the first power consumption demand index, retain the first initial energy supply plan and calculate the first new energy power generation ratio. Specifically, when evaluating the first initial energy supply plan, first determine the specific values of the first overall power supply reliability coefficient and the first power consumption demand index. The former is calculated from the ratios and historical reliability coefficients of each power supply method in the plan, such as 0.8, and the latter is obtained by considering factors such as production and environmental intensity, assumed to be 0.75. Compare the two. When the first overall power supply reliability coefficient is greater than or equal to the first power consumption demand index, in this example, 0.8 ≥ 0.75, the plan meets the requirements and is retained, and the plan contains the ratio information of various power supply methods. Then, clarify that photovoltaic, wind power, etc. are new energy power supply methods, and extract their ratios from the retained plan. For example, photovoltaic is 0.3 and wind power is 0.4, and add them up to calculate the first new energy power generation ratio as 0.7. This ratio reflects the proportion of new energy in the power supply plan under the condition of meeting the reliability of power consumption demand, which is of great significance for evaluating the energy structure and sustainability.

[0048] Step S335, if the first overall power supply reliability coefficient is less than the first power consumption demand index, it is discarded. Specifically, in the evaluation of the energy supply plan, it is necessary to first clarify the first overall power supply reliability coefficient and the first power consumption demand index. The former, for example, a certain plan includes photovoltaic, wind power, and energy storage, with proportions of 0.3, 0.4, and 0.3 respectively, and the corresponding historical reliability coefficients are 0.6, 0.7, and 0.8. After weighted calculation, it is 0.69; the latter comprehensively considers production and environmental intensity factors and is set to 0.75. Comparing the two, in this example, 0.69 is less than 0.75, indicating that this first initial energy supply plan cannot meet the power supply reliability requirements and is immediately discarded. After discarding, it is necessary to regenerate the initial energy supply plan, randomly generate the energy supply proportion again, and combine it with the predicted energy output power set, repeating the entire process until a plan that meets the requirements is found.

[0049] Step S336, perform iterative selection, screening, and calculation of the new energy power generation proportion until the predetermined number of selections is reached, and output the energy supply proportion corresponding to the maximum new energy power generation proportion as the first predicted energy supply proportion. Specifically, in the optimization of the energy supply plan, first set the predetermined number of selections, such as 100 times, and initialize the variables used to record the new energy power generation proportion and the corresponding energy supply proportion. Each iteration aims to meet the average value of the first predicted energy consumption power. Randomly generate an initial energy supply plan according to the predicted energy output power set. Assume that the power supply proportions of photovoltaic, wind power, and energy storage are 0.25, 0.35, and 0.4 respectively. Then, using the energy supply proportion of this plan as the weight, calculate the first overall power supply reliability coefficient in combination with the historical power supply reliability coefficient, such as 0.7325. Compare it with the first power consumption demand index (set to 0.7). If the requirement is met, retain the plan and calculate the new energy power generation proportion (0.6 in this example), and record the relevant data; if not, discard it. Repeat this process until the predetermined number of selections is reached. After the end, find the maximum new energy power generation proportion in the records, such as 0.7, and its corresponding energy supply proportion, and output it as the first predicted energy supply proportion, providing a key basis for energy planning decisions.

[0050] Step S400: Calculate the carbon emissions for the predetermined future time zone based on the unit energy carbon emission factor, the predicted energy consumption sequence, and the predicted energy supply ratio sequence, and obtain the predicted carbon emissions as the result of carbon emission trend deduction. Specifically, when calculating the carbon emissions for the predetermined future time zone, data needs to be comprehensively collected first. Identify the unit energy carbon emission factors corresponding to different power supply methods. For example, assume that for thermal power it is 0.8 tons of carbon dioxide per megawatt-hour, for hydropower it is 0.05 tons of carbon dioxide per megawatt-hour, for wind power it is 0.02 tons of carbon dioxide per megawatt-hour, etc. These data can be obtained from energy industry reports, etc. At the same time, obtain the predicted energy consumption sequence. For example, at hourly intervals, the predicted values for the next 24 hours are like [100, 120, 110, …, 90] megawatt-hours. Also obtain the predicted energy supply ratio sequence obtained through previous iterative screening, which is assumed to be presented in the form of a two-dimensional array showing the proportions of each power supply method. Then perform the carbon emissions calculation. For each time node, calculate the energy supply volume of each power supply method according to the predicted energy consumption and supply ratio, and then calculate the carbon emissions in combination with its unit energy carbon emission factor. Accumulate to obtain the total carbon emissions at this time node, and then generate a predicted carbon emissions sequence. Finally, output this sequence as the result of carbon emission trend deduction, which can be displayed in the form of charts or data tables to provide a decision-making basis for energy planning and carbon emission management.

[0051] In the embodiments of the present application, by combining the production plan and climate characteristics of the industrial park in the future period, the energy consumption sequence for continuous P time points is predicted, and at the same time, the output power sequence of multi-source energy is predicted based on historical power generation records; on this basis, the energy supply proportion is analyzed and a supply ratio sequence is generated. Finally, the unit energy carbon emission factor is used to integrate the energy consumption sequence and the supply ratio sequence to calculate the carbon emissions in the future time zone, realizing the accurate deduction of the carbon emission trend, achieving the technical effects of reducing the carbon emission prediction error rate and improving the dynamic optimization response speed of the new energy supply proportion.

[0052] In the above text, with reference to Figure 1 The digital park carbon emission trend deduction method according to the embodiments of the present invention is described in detail. Next, with reference to Figure 2 The digital park carbon emission trend deduction system according to the embodiments of the present invention will be described.

[0053] The digital park carbon emission trend deduction system according to the embodiments of the present invention is used to solve the technical problems of insufficient multi-source energy collaborative modeling and lagging dynamic power supply ratio optimization in the existing park carbon emission prediction, and achieves the technical effects of reducing the carbon emission prediction error rate and improving the dynamic optimization response speed of the new energy supply proportion. The digital park carbon emission trend deduction system includes: a predicted energy consumption sequence generation module 10, a predicted energy output power sequence generation module 20, an energy supply proportion analysis module 30, and a carbon emissions calculation module 40.

[0054] The predicted energy consumption sequence generation module 10 is used to perform energy consumption prediction for consecutive P time points according to the production plan and climate characteristics of the industrial park in a predetermined future time zone, and generate a predicted energy consumption sequence.

[0055] The predicted energy output power sequence generation module 20 is used to perform multi-energy output power prediction for consecutive P time points according to historical power generation records, and generate multiple predicted energy output power sequences.

[0056] The energy power supply ratio analysis module 30 is used to perform energy power supply ratio analysis based on the predicted energy consumption sequence and multiple predicted energy output power sequences, and generate a predicted energy power supply ratio sequence.

[0057] The carbon emission calculation module 40 is used to calculate the carbon emissions in the predetermined future time zone according to the unit energy carbon emission factor, the predicted energy consumption sequence and the predicted energy power supply ratio sequence, and obtain the predicted carbon emissions as the carbon emission trend deduction result.

[0058] Next, the specific configuration of the predicted energy consumption sequence generation module 10 will be described in detail. As described above, according to the production plan and climate characteristics of the industrial park in a predetermined future time zone, energy consumption prediction is performed for consecutive P time points to generate a predicted energy consumption sequence. The predicted energy consumption sequence generation module 10 further includes: a climate characteristic acquisition unit for acquiring the production plan and climate characteristics of the industrial park in a predetermined future time zone, where the production plan is the expected product output and the climate characteristic is the expected environmental temperature; a time interval division unit for setting consecutive P time points in the predetermined future time zone at a predetermined time interval and dividing to obtain P time intervals, where P is an integer greater than 30; an energy consumption correlation data determination unit for dividing and combining the production plan and climate characteristics according to the P time intervals to determine P energy consumption correlation data; a power mean calculation unit for using an energy consumption prediction plug-in to perform predictions respectively according to the P energy consumption correlation data, outputting P predicted energy consumption amounts, calculating P predicted energy consumption power means, and constructing the predicted energy consumption sequence.

[0059] Among them, using the energy consumption prediction plug-in, predictions are respectively made according to the P energy consumption correlation data, P predicted energy consumption quantities are output, the average value of the P predicted energy consumption powers is calculated, and the predicted energy consumption sequence is constructed. The power average value calculation unit further includes: an associated data set acquisition subunit, which is used to collect a sample energy consumption associated data set based on the historical production records of the industrial park, and count the energy consumption quantities corresponding to different energy consumption associated data according to the historical electricity consumption records, set as sample energy consumption quantities, and construct a sample energy consumption quantity set; an energy consumption prediction branch acquisition subunit, which is used to use the sample energy consumption associated data set and the sample energy consumption quantity set as training data, and train multiple prediction operators respectively to obtain multiple energy consumption prediction branches; an energy consumption prediction plug-in construction subunit, which is used to integrally construct the energy consumption prediction plug-in according to the multiple energy consumption prediction branches. Among them, the output of the energy consumption prediction plug-in is the average value of the outputs of the multiple energy consumption prediction branches.

[0060] Among them, using the sample energy consumption associated data set and the sample energy consumption quantity set as training data, multiple prediction operators are trained respectively to obtain multiple energy consumption prediction branches. The energy consumption prediction branch acquisition subunit further includes: a prediction operator configuration micro-unit, which is used to configure K prediction operators based on machine learning, where K is greater than or equal to 3; a loss function convergence micro-unit, which is used to divide the training data into K equal parts, and respectively perform supervised training and verification on the K prediction operators until the loss function converges to obtain multiple energy consumption prediction branches.

[0061] Next, the specific configuration of the predicted energy output power sequence generation module 20 will be described in detail. As described above, according to the historical power generation records, multi-energy output power predictions are made at P consecutive time points to generate multiple predicted energy output power sequences. The predicted energy output power sequence generation module 20 further includes: a power supply method acquisition unit, which is used to acquire the multi-source power supply method of the industrial park and multiple historical power generation records of the multi-source power supply method; a power generation impact data set acquisition unit, which is used to acquire P power generation impact data sets of the multi-source power supply method within P time intervals; an integrated power predictor construction unit, which is used to construct an integrated power predictor based on machine learning according to the historical power generation records, and respectively perform multi-energy output power predictions according to the P power generation impact data sets to generate multiple predicted energy output power sequences.

[0062] Next, the specific configuration of the energy supply ratio analysis module 30 will be described in detail. As described above, based on the predicted energy consumption sequence and multiple predicted energy output power sequences, energy supply ratio analysis is performed to generate a predicted energy supply ratio sequence. The energy supply ratio analysis module 30 further includes: a first time point selection unit for selecting the mean value of the first predicted energy consumption power and the set of first predicted energy output powers at the first time point; a demand index configuration unit for configuring the first electricity demand index according to the first product output and the first environmental temperature in the first time interval; and a first predicted energy supply ratio output unit for performing energy supply ratio analysis based on the mean value of the first predicted energy consumption power and the set of first predicted energy output powers with the first electricity demand index as a constraint and the maximum proportion of new energy power generation as a target, and outputting the first predicted energy supply ratio, which is added to the predicted energy supply ratio sequence.

[0063] Among them, when configuring the first electricity demand index according to the first product output and the first environmental temperature in the first time interval, the demand index configuration unit further includes: a product output mean ratio calculation subunit for calculating the ratio of the first product output to the mean value of the product outputs in the historical time interval, which is set as the first production intensity; a first environmental intensity setting subunit for calculating the ratio of the first environmental temperature to the mean value of the environmental temperatures in the historical time interval, which is set as the first environmental intensity; and a first electricity intensity calculation subunit for calculating the first electricity intensity based on the first production intensity and the first environmental intensity, and setting the first power supply reliability coefficient according to the first electricity intensity as the first electricity demand index, where the power supply reliability coefficient is positively correlated with the electricity intensity.

[0064] Among them, with the first power consumption demand index as a constraint and the maximum proportion of new energy power generation as the goal, energy supply ratio analysis is carried out according to the first predicted average energy consumption power and the first predicted energy output power set, and the first predicted energy supply ratio is output and added to the predicted energy supply ratio sequence. The first predicted energy supply ratio output unit further includes: a stable power supply ratio statistics subunit, which is used to respectively count the stable power supply ratios of each power supply method in the multi-source power supply method within a predetermined historical time period according to historical power generation records, set as historical power supply reliability coefficients, and obtain a plurality of historical power supply reliability coefficients; a first initial energy supply plan generation subunit, which is used to generate a first initial energy supply plan randomly according to the predicted energy output power set with the goal of meeting the first predicted average energy consumption power, where the initial energy supply plan includes the energy supply ratio; a reliability coefficient calculation subunit, which is used to calculate the first overall power supply reliability coefficient of the first initial energy supply plan based on the plurality of historical power supply reliability coefficients with the energy supply ratio as the weight; a first new energy power generation ratio calculation subunit, which is used to retain the first initial energy supply plan and calculate the first new energy power generation ratio if the first overall power supply reliability coefficient is greater than or equal to the first power consumption demand index; a power demand index determination subunit, which is used to discard if the first overall power supply reliability coefficient is less than the first power consumption demand index; a predetermined selection times judgment subunit, which is used to perform iterative selection, screening, and new energy power generation ratio calculation until the predetermined selection times are reached, and output the energy supply ratio corresponding to the maximum new energy power generation ratio as the first predicted energy supply ratio.

[0065] The digital park carbon emission trend deduction system provided by the embodiments of the present invention can execute the digital park carbon emission trend deduction method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0066] Based on the foregoing embodiments, the embodiments of the present application also provide an electronic device, and when the processor of the electronic device is executed, it can implement the method described in any previous embodiment.

[0067] Figure 3 It is a schematic structural diagram of the electronic device provided by the embodiments of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention. The electronic device is in the form of a general computing device, and its components may include but are not limited to an input device 201, a processor 202, a memory 203, and an output device 204. Among them, the processor 202 may be one or more; the processor 202 executes various functional applications and data processing of the computer device by running the software programs, instructions, and modules stored in the memory 203, that is, realizing the above-mentioned digital park carbon emission trend deduction method.

[0068] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0069] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A digital park carbon emission trend deduction method, characterized in that: Methods include: According to the production plan and climate characteristics of the industrial park in the predetermined future time zone, the energy consumption at P consecutive time points is predicted to generate a predicted energy consumption sequence; According to the historical power generation records, multiple energy output power forecasts are carried out at P consecutive time points to generate multiple predicted energy output power sequences; Based on the predicted energy consumption sequence and multiple predicted energy output power sequences, energy power supply ratio analysis is performed to generate a predicted energy power supply ratio sequence; Calculating the carbon emissions in the predetermined future time zone according to the unit energy carbon emission factor, the predicted energy consumption sequence, and the predicted energy power supply ratio sequence, and obtaining the predicted carbon emissions as a carbon emission trend deduction result, including: First, we need to collect data comprehensively and clarify the carbon emission factors per unit of energy corresponding to different power supply methods; At the same time, the energy consumption series and the predicted energy supply ratio series obtained through previous iterative screening are obtained; Next, the carbon emissions are calculated. For each time node, the energy supply of each power supply mode is calculated based on the predicted energy consumption and power supply ratio, and then the carbon emissions are calculated by combining the unit energy carbon emission factor. The total carbon emissions at the time node are accumulated to generate a predicted carbon emissions sequence. Finally, this sequence is output as the result of carbon emission trend deduction and can be displayed in charts or data tables to provide a decision-making basis for energy planning and carbon emission management.

2. The digital park carbon emission trend deduction method according to claim 1 is characterized in that: According to the production plan and climate characteristics of the industrial park in the predetermined future time zone, the energy consumption at P consecutive time points is predicted to generate a predicted energy consumption sequence, including: Obtain the production plan and climate characteristics of the industrial park in a predetermined future time zone, where the production plan is the expected product output and the climate characteristics are the expected ambient temperature; Setting P consecutive time points in the predetermined future time zone at predetermined time intervals, and dividing and obtaining P time intervals, where P is an integer greater than 30; Dividing and combining the production plan and climate characteristics according to the P time intervals to determine P energy consumption related data; By using the energy consumption prediction plug-in, predictions are made according to the P energy consumption related data respectively, P predicted energy consumption amounts are output, the P predicted energy consumption power means are calculated, and the predicted energy consumption sequence is constructed.

3. The digital park carbon emission trend deduction method according to claim 2 is characterized in that: Build an energy consumption forecast plugin, including: Based on the historical production records of the industrial park, a sample energy consumption related data set is collected, and the energy consumption corresponding to different energy consumption related data is counted according to the historical electricity consumption records, which is set as the sample energy consumption, and a sample energy consumption set is constructed; The sample energy consumption associated data set and the sample energy consumption amount set are used as training data to train multiple prediction operators respectively, and multiple energy consumption prediction branches are obtained; The energy consumption prediction plug-in is constructed based on the integration of the multiple energy consumption prediction branches, wherein the output of the energy consumption prediction plug-in is the average of the outputs of the multiple energy consumption prediction branches.

4. The digital park carbon emission trend deduction method according to claim 3 is characterized in that: Train multiple prediction operators separately to obtain multiple energy consumption prediction branches, including: Configure K prediction operators based on machine learning, where K is greater than or equal to 3; The training data is equally divided into K parts, and supervised training and verification are performed on the K prediction operators respectively until the loss function converges, thereby obtaining a plurality of energy consumption prediction branches.

5. The digital park carbon emission trend deduction method according to claim 2 is characterized in that: According to the historical power generation records, the multi-energy output power forecast is carried out at P consecutive time points to generate multiple predicted energy output power sequences, including: Obtain the multi-source power supply mode of the industrial park and multiple historical power generation records of the multi-source power supply mode; Acquire P power generation impact data sets of the multi-source power supply mode within P time intervals; According to historical power generation records, an integrated power predictor is constructed based on machine learning, and multi-energy output power predictions are performed separately according to the P power generation impact data sets to generate multiple predicted energy output power sequences.

6. The method for deriving carbon emission trends of a digital park according to claim 2 is characterized in that: Based on the predicted energy consumption sequence and multiple predicted energy output power sequences, energy power supply ratio analysis is performed to generate a predicted energy power supply ratio sequence, including: Selecting a first predicted energy consumption power mean value and a first predicted energy output power set at a first time point; configuring a first electricity demand indicator according to a first product output and a first ambient temperature in a first time interval; Taking the first electricity demand indicator as a constraint and the maximum proportion of new energy power generation as the goal, an energy supply ratio analysis is performed based on the first predicted energy consumption power mean and the first predicted energy output power set, and the first predicted energy supply ratio is output and added to the predicted energy supply ratio sequence.

7. The method for deriving carbon emission trends of a digital park according to claim 6 is characterized in that: Configuring a first power demand indicator according to a first product output and a first ambient temperature in a first time interval includes: Calculate the ratio of the output of the first product to the average output of the product in the historical time interval, and set it as the first production intensity; Calculate the ratio of the first ambient temperature to the average ambient temperature in a historical time interval, and set it as the first ambient intensity; A first power consumption intensity is calculated according to the first production intensity and the first environmental intensity, and a first power supply reliability coefficient is set according to the first power consumption intensity as a first power demand indicator, wherein the power supply reliability coefficient is positively correlated with the power consumption intensity.

8. The method for deriving carbon emission trends of a digital park according to claim 7 is characterized in that: Output the first predicted energy supply ratio, including: According to the historical power generation records, the stable power supply ratio of each power supply mode in the multi-source power supply mode within the predetermined historical time zone is counted respectively, which is set as the historical power supply reliability coefficient, and multiple historical power supply reliability coefficients are obtained; With the first predicted energy consumption power mean as the goal, a first initial energy power supply plan is randomly generated according to the predicted energy output power set, wherein the initial energy power supply plan includes an energy power supply ratio; Taking the energy power supply proportion as a weight and based on the multiple historical power supply reliability coefficients, a first overall power supply reliability coefficient of the first initial energy power supply scheme is calculated; If the first overall power supply reliability coefficient is greater than or equal to the first power demand index, the first initial energy power supply plan is retained, and the first new energy power generation ratio is calculated; If the first overall power supply reliability coefficient is less than the first power demand index, then discard it; Iterative selection, screening and calculation of the proportion of new energy power generation are performed until the predetermined number of selections is reached, and the energy supply proportion corresponding to the maximum proportion of new energy power generation is output as the first predicted energy supply proportion.

9. The digital park carbon emission trend deduction system is characterized by: The system is used to implement the digital park carbon emission trend deduction method according to any one of claims 1 to 8, and the system includes: A predicted energy consumption sequence generation module, wherein the predicted energy consumption sequence generation module is used to perform energy consumption forecasting at P consecutive time points according to the production plan and climate characteristics of the industrial park in a predetermined future time zone, and generate a predicted energy consumption sequence; A predicted energy output power sequence generation module, wherein the predicted energy output power sequence generation module is used to perform multi-energy output power prediction at P consecutive time points according to historical power generation records, and generate multiple predicted energy output power sequences; An energy supply ratio analysis module, the energy supply ratio analysis module is used to perform energy supply ratio analysis based on the predicted energy consumption sequence and multiple predicted energy output power sequences, and generate a predicted energy supply ratio sequence; A carbon emission calculation module is used to calculate the carbon emissions in the predetermined future time zone according to the unit energy carbon emission factor, the predicted energy consumption sequence and the predicted energy power supply ratio sequence, and obtain the predicted carbon emissions as the carbon emission trend deduction result.

10. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable instructions; The processor is used to implement the digital park carbon emission trend deduction method described in any one of claims 1 to 8 when executing the executable instructions stored in the memory.

Images