Typical low-carbon project carbon reduction investment intensity measuring and calculating system

Abstract

The invention relates to the technical field of intelligent management of carbon emission of low-carbon engineering, and discloses a typical low-carbon engineering carbon reduction investment intensity measuring and calculating system, which acquires operation data of low-carbon engineering such as offshore photovoltaic engineering through multi-sensor redundancy, ensures credible storage of the data by utilizing a block chain technology, and predicts environment and market factors by adopting a deep learning model. The system power is dynamically distributed based on a multi-objective optimization algorithm, the carbon reduction investment intensity is automatically calculated in combination with the investment cost and the carbon emission reduction amount, and visual presentation and intelligent early warning are achieved through the digital twinborn technology. According to the method, the technical problems that the existing low-carbon engineering carbon reduction benefit evaluation is inaccurate and the investment income is difficult to quantify are solved, and the scientificity and the accuracy of low-carbon engineering investment decision making are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management of carbon emissions from low-carbon projects, and more specifically, to a system for calculating the carbon reduction investment intensity of typical low-carbon projects. Background Art

[0002] Low-carbon projects such as offshore photovoltaic power generation and seawater desalination have been widely adopted in recent years as important technological pathways for reducing carbon emissions. However, the construction of low-carbon projects often requires significant capital investment, and evaluating the relationship between the carbon reduction benefits and investment costs of these projects has become a key issue for decision makers.

[0003] In existing technologies, the carbon reduction benefit assessment of low-carbon projects mainly relies on simplified static calculation methods, which cannot fully consider dynamic influencing factors such as environmental fluctuations, market price changes, and equipment efficiency degradation, resulting in a large deviation between the measurement results and the actual operating effects. In addition, traditional evaluation methods generally have technical defects such as low data credibility, poor prediction accuracy, single optimization strategy, and incomplete evaluation dimensions. For example, the power scheduling of the current offshore photovoltaic and seawater desalination system mostly adopts fixed rules or manual experience, which is difficult to adapt to the complex and changing marine environment; the calculation of carbon reduction is usually based on theoretical values ​​or static parameters, which fails to reflect the actual operating status; and the investment cost accounting often ignores factors such as equipment performance changes and market fluctuations.

[0004] Therefore, there is an urgent need for a technical solution that can collect operating data in real time, ensure data credibility, accurately predict environmental changes, dynamically optimize power distribution, and scientifically calculate carbon reduction investment intensity, so as to solve the technical problems of inaccurate carbon reduction benefit assessment of low-carbon projects and difficulty in quantifying investment returns. Summary of the Invention

[0005] The present invention provides a typical low-carbon project carbon reduction investment intensity calculation system to solve technical problems in related technologies such as inaccurate carbon reduction benefit assessment of low-carbon projects and difficulty in quantifying benefits.

[0006] The present invention provides a typical low-carbon project carbon reduction investment intensity calculation system, including: Data acquisition and preprocessing module, used to collect raw data from low-carbon project sites and perform preprocessing; Blockchain trusted storage module, used to ensure the integrity and traceability of original data and prevent data tampering; The model prediction module predicts future environmental and market factors based on raw data, providing a decision basis for dynamic power allocation; Dynamic power allocation module, used to dynamically optimize system power allocation based on model prediction results to maximize carbon reduction benefits; The carbon emission reduction benefit assessment module is used to calculate the ratio of carbon emission reduction to investment cost of low-carbon projects in real time and quantify the carbon reduction investment intensity index; Intelligent early warning and decision support module, used to monitor system operation status, identify potential problems and provide decision-making suggestions.

[0007] In a preferred embodiment, the data acquisition and preprocessing module includes: Multi-sensor redundant data collection, including active and standby sensor pairs, is used to alternately collect information on photovoltaic output, wave height, wave period, sunshine intensity, seawater salinity, campus load, and electricity and carbon prices; Multi-link transmission fusion, including submarine cable, satellite, and 4G / 5G transmission links, automatically selects the optimal transmission path through a weight function; Time synchronization, using GPS and PTP dual time calibration mechanism to ensure the time consistency of each collection point; Unified format: data from different sources and formats are converted into standard JSON format through a protocol gateway; Missing and abnormal data processing uses sliding window statistics combined with machine learning to fill in missing data and correct abnormal data.

[0008] In a preferred embodiment, the blockchain trusted storage module includes: Data hashing: Use the SHA-256 algorithm to calculate the hash value of the time series data set and call the smart contract to upload the hash value to the blockchain network; Consensus nodes use a consortium chain architecture and select PBFT or Raft consensus algorithms based on the characteristics of the deployment environment to ensure that all nodes reach consensus on the data hash records; The on-chain index uses a B+ tree structure to organize the timestamp index, maintains a mapping table from hash value to transaction ID, and supports efficient time range queries and hash value queries.

[0009] In a preferred embodiment, the model prediction module includes: Federated learning training uses the FedAvg algorithm to collaboratively train the prediction model on multiple nodes. The prediction model structure adopts a deep learning architecture that combines LSTM with an attention mechanism. Model compression deployment: deploying the model to edge computing devices through network pruning and parameter quantization; Online prediction: trigger a prediction task every hour, input the cleaned raw data of the past 168 hours, and generate a prediction sequence for the next 24 hours; Interpretable output, using the shape additive interpretation method to analyze the factor contribution of each predicted value; Adaptive fine-tuning triggers model fine-tuning based on the mean absolute percentage error (MAPE).

[0010] In a preferred embodiment, the dynamic power allocation module includes: Initial solution generation: quickly generate feasible initial dispatch plans based on PV power generation priority, desalination load matching, electricity price response, carbon price response, and safety margin rules; Deep optimization uses an improved particle swarm optimization algorithm or actor-critic reinforcement learning algorithm to generate a better scheduling solution. The optimization objective function is: ; in, represents the optimization objective function value, represents the time step, represents the carbon cost weight, represents the economic benefit weight, represents the carbon emission reduction per unit water production at time t, represents the carbon emission reduction at time t, represents the water production at time t, represents the economic benefits at time t; Digital twin verification: importing the optimized candidate solutions into the digital twin model for simulation verification; Failover: Automatically switch to a predefined fixed-mode backup strategy in the event of device anomalies or communication interruptions.

[0011] In a preferred embodiment, the carbon emission reduction benefit assessment module includes: Carbon emission reduction calculation is used to calculate the daily cumulative carbon reduction of the system. The calculation formula is: ; in, represents the daily cumulative carbon reduction, represents the optimized photovoltaic power at time t, represents the carbon emission reduction factor of photovoltaic power generation; represents the water production at time t, represents the carbon emission factor for seawater desalination; Investment cost calculation, calculate the total investment cost of the system, including equipment purchase cost, installation and commissioning cost, and operation and maintenance cost; The investment intensity index is generated and the carbon reduction investment intensity index is calculated based on the daily cumulative carbon reduction amount and the total investment cost.

[0012] In a preferred embodiment, the carbon reduction investment intensity formula calculated by the investment intensity index generation unit is: ; in, represents the daily cumulative carbon reduction, represents the total investment cost, It represents the intensity of carbon reduction investment, with the unit being tons of carbon dioxide per 10,000 yuan per year.

[0013] In a preferred embodiment, the intelligent early warning and decision support module includes: Threshold monitoring: regularly scanning system operation data to identify carbon reduction discrepancies, abnormal investment intensity, and abnormal price fluctuations; Optimization suggestion: call the dynamic power allocation module to regenerate the scheduling plan according to the exception type; Notification push: push warning information and optimization suggestions to relevant personnel via SMS, email, and mobile push channels.

[0014] In a preferred embodiment, the carbon reduction difference calculated by the threshold monitoring unit is: ; in, represents the carbon reduction difference, Indicates the actual carbon reduction amount, Indicates the planned carbon reduction amount.

[0015] In a preferred embodiment, a computer-readable storage medium is used to store computer-readable instructions, which, when read by a computer, can run the above-mentioned typical low-carbon project carbon reduction investment intensity calculation system.

[0016] The beneficial effects of the present invention are: Through multi-sensor redundant data collection, multi-link transmission fusion, blockchain trusted storage and other technical means, the accuracy and reliability of data are improved, providing a reliable data foundation for the calculation of carbon reduction investment intensity; A deep learning model trained with federated learning enables accurate prediction of environmental and market factors. Combined with a multi-objective optimization algorithm, it dynamically allocates system power, significantly improving system operating efficiency and carbon reduction benefits. A complete investment intensity calculation method has been established, which comprehensively considers multiple factors such as carbon emission reduction and investment cost, and achieves scientific quantification of the benefits of carbon reduction investment; Digital twin technology enables visualization of system operating status, and intelligent early warning mechanisms provide decision support, effectively enhancing the system's operability and decision-making support capabilities. The system's scalability and operational efficiency are improved through distributed architecture and heterogeneous computing; data transmission latency and storage pressure are reduced with the help of edge computing technology; the AI ​​model based on adaptive fine-tuning ensures long-term prediction accuracy; resource allocation efficiency is improved through data-driven dynamic optimization; the system's fault tolerance and reliability are enhanced by using intelligent early warning mechanisms; and the human-computer interaction experience is improved by using immersive visualization technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a module diagram of a typical low-carbon project carbon reduction investment intensity calculation system of the present invention. DETAILED DESCRIPTION

[0018] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses a typical low-carbon project carbon reduction investment intensity calculation system, such as Figure 1 Shown, including: Data acquisition and preprocessing module, used to collect raw data from low-carbon project sites and perform preprocessing; In one embodiment of the present invention, the raw data stream mainly includes information such as photovoltaic output, wave height, wave period, sunshine intensity, seawater salinity, park load, and electricity and carbon prices. The specific data collection and preprocessing includes the following steps: S101, multi-sensor redundant acquisition; The system achieves redundant data collection through sensor groups. These groups include primary and backup sensor pairs. For example, in the case of photovoltaic output sensors, the system deploys two independent photovoltaic output monitoring devices, designated as primary and backup sensors. These sensors alternately collect data at a preset sampling frequency (e.g., every 5 minutes). When a primary sensor failure is detected (e.g., three consecutive abnormal sampling values ​​or no response), the system automatically switches to the backup sensor and issues a maintenance reminder for the primary sensor. This redundant acquisition mechanism ensures the continuity of data acquisition in the event of a single point of failure. S102, multi-link transmission fusion; The collected raw data is transmitted through a data transmission network that includes three transmission links: submarine cables, satellites, and 4G / 5G. The system monitors the transmission quality indicators of each link (including latency, packet loss rate, and bandwidth) in real time and automatically selects the optimal transmission path based on a preset weight function. The weight function can be expressed as: ; in, represents the comprehensive weight of the i-th link, represents the normalized delay of the i-th link, represents the normalized packet loss rate of the i-th link, represents the normalized bandwidth of the i-th link, 、 and Represent the delay, packet loss, and bandwidth weight coefficients respectively, and satisfy ;choose The smallest link is used as the current optimal transmission path (i.e., the path with the highest transmission efficiency).

[0020] S103, time synchronization; A dual time synchronization mechanism using GPS and PTP (Precision Time Protocol) ensures time consistency at all collection points. Each collection point is equipped with a GPS receiver as the primary time source, while PTP is run over the network as a backup time synchronization mechanism. The difference between the two time sources is regularly compared. When the difference exceeds a preset threshold (e.g., 5ms), an alarm is triggered and manual intervention is initiated. Through this dual time synchronization mechanism, the system ensures that time synchronization errors are controlled within 10ms. S104, unified format; The protocol gateway converts data from different sources and formats into a standard JSON format. The protocol gateway supports mainstream industrial protocols such as Modbus, OPC UA, and MQTT, and maps data points from various protocols to unified JSON fields based on pre-configured mapping tables. Taking the Modbus protocol as an example, the gateway maps the Modbus register address to a specific JSON field name and performs appropriate conversion based on the data type; the unified JSON data contains necessary metadata (such as timestamp, source identifier, quality code, etc.) and the actual measurement value.

[0021] S105, missing and exception handling; Perform missing data detection and anomaly identification on unified formatted data; the system uses sliding window statistics combined with machine learning to check data quality; Specifically, the system maintains a historical data window of length n (for example, n=24, representing 24 hours) and calculates the statistical characteristics (mean, standard deviation, rate of change, etc.) of the data within the window. For newly collected data points, the system calculates their anomaly scores using the pre-trained isolation forest algorithm: ; in, represents the anomaly score of sample x, represents the average path length of sample x, represents the expected path length, is the normalization factor, and n is the window length. When it exceeds the preset threshold (for example, 0.7), it is determined to be an outlier.

[0022] For detected missing data, the system uses different filling strategies based on the data type: For slowly varying data (such as seawater salinity), linear interpolation is used to fill in the gaps; For periodic data (such as photovoltaic output), the historical average value of the same period is used to fill the gap; For random data (such as electricity prices), the previous valid value is used to fill in the gaps; For abnormal data detected, the system first attempts to switch to the backup sensor data; if the backup data is also abnormal, it starts the online calibration program to correct the abnormal value based on the historical data model; The specific correction steps are as follows: Build a historical data model. The system builds an ARIMA time series forecasting model based on the valid original data of the same period in the past 30 days. For abnormality assessment, the normalized root mean square error (NRMSE) method is used to calculate the deviation between the abnormal value and the model prediction value, and the correction weight is determined according to the deviation size; Weighted correction: the outliers and the predicted values ​​of the historical data model are weighted averaged according to the calculated weights to generate the corrected data values; Confidence mark: add confidence mark to the corrected data for reference by subsequent modules; The process of building a historical data model includes: Data segmentation: categorizes low-carbon project site data such as photovoltaic output, wave height / period, and sunshine intensity according to time periods and working / non-working days. Different segmentation strategies are used for slowly varying, periodic, and random data. Feature extraction: extracting the trend, periodicity, and autocorrelation characteristics of various sensor data, with particular attention to the correlation between photovoltaic output and sunshine intensity, and seasonal changes in seawater salinity; Parameter optimization, determining the optimal parameters (p, d, q) of the ARIMA model (autoregressive integrated moving average model, used for analyzing and forecasting time series data) through grid search (an exhaustive parameter optimization technique that systematically tries all possible parameter combinations in a predefined parameter space and evaluates the model performance for each set of parameters); Where p is the number of autoregressive terms (the degree to which past values ​​influence the current value), d is the order of differencing (the number of differences required to transform a non-stationary time series into a stationary series), and q is the number of moving average terms (the degree to which past forecast errors influence the current value). Differentiated parameter settings are used for different types of sensor data. For example, photovoltaic output data may require higher-order differential processing (d value) to eliminate seasonal fluctuations and trend changes, wave height data may require more autoregressive terms (p value) to capture its cyclical variation patterns, and electricity price data may require more moving average terms (q value) to smooth random fluctuations. Model validation: k-fold cross-validation (k=5) was performed using data from the same period over the past 30 days. The dataset was divided into five subsets, and the model was trained using four subsets each time and tested on the remaining subset. After five cycles, the average performance index was taken to evaluate the model's prediction accuracy in different time windows to ensure the model's stability and reliability during the abnormal data correction process. The data acquisition and preprocessing module outputs a high-quality time series dataset that has undergone multi-sensor redundant acquisition, multi-link transmission fusion, time synchronization, format unification, and missing and anomaly processing. The dataset contains the following key fields: Data collection timestamp is accurate to millisecond level and synchronized by GPS and PTP dual time synchronization mechanism; PV system output power, after anomaly detection and correction; Wave height, used to assess wave energy generation potential; The wave period, together with the wave height, determines the efficiency of wave energy conversion; The intensity of sunlight directly affects the efficiency of photovoltaic power generation; Seawater salinity affects desalination energy consumption and equipment life; The park's electricity load reflects the energy demand side; electricity prices, reflecting energy market conditions; Carbon price, used for carbon emission cost accounting; Each data record is accompanied by a quality code, indicating whether the data is the original collected value, interpolated value or model-corrected value, as well as the corresponding confidence score.

[0023] Blockchain trusted storage module, used to ensure the integrity and traceability of original data and prevent data tampering; In one embodiment of the present invention, the processing process of the blockchain trusted storage module includes the following steps: S201, data hashing and packaging; The data hash unit performs hash processing on the received time series data set; Every hour, the system aggregates the time series data set within one hour into a data block and calculates the hash value of the data block. The hash algorithm uses SHA-256 to ensure data integrity and immutability. The calculation formula is: in, represents the timestamp of the i-th data point, represents the data content of the i-th data point, n represents the number of data points, Represents a string concatenation operation, and n represents the total number of data points.

[0024] After calculating the hash value, the data hash unit calls the smart contract uploadData(hash, timestamp) to upload the hash value and the corresponding timestamp to the blockchain network. uploadData is a smart contract function used to upload data to the blockchain network for storage and verification. The basic logic of smart contracts includes permission verification, data format verification, and storage operations, ensuring that only authorized nodes can upload data and that the data format complies with predefined specifications. S202, alliance chain consensus; Consensus nodes participate in the consensus process of the blockchain network. The system adopts a consortium chain architecture, including multiple distributed consensus nodes. Depending on the characteristics of the deployment environment, the system can choose PBFT (Practical Byzantine Fault Tolerance) or Raft consensus algorithm. For scenarios requiring higher security, the system uses the PBFT algorithm, which can ensure system consistency even when up to one-third of the nodes experience Byzantine failures (i.e., nodes may send incorrect information or behave maliciously). The PBFT consensus process consists of four phases: request, pre-prepare, prepare, and confirm. A three-phase commit protocol ensures that all honest nodes reach consensus. Optionally, for scenarios that prioritize performance, the system uses the Raft algorithm, which works under the assumption that nodes are honest but may fail, and has high throughput. Raft achieves distributed consensus through leader election, log replication, and safety rules. Through these consensus mechanisms, the system ensures that all nodes reach a consensus on the data hash records, preventing single point nodes from tampering with data; S203, on-chain index; On-chain indexing creates and maintains data indexes on the blockchain. Whenever a new data hash is confirmed on the chain, the system records the transaction's block number, transaction ID, data hash value, and corresponding timestamp, forming an index record {block number, transaction ID, data hash value, timestamp}. These index records are stored in a dedicated index data structure, supporting efficient time range queries and hash value queries. To improve query efficiency, the system uses a B+ tree structure to organize timestamp indexes. The B+ tree structure is a balanced binary tree with efficient search performance, making it particularly suitable for range queries and time series data. It also maintains a mapping table from hash values ​​to transaction IDs. This allows the system to quickly locate data records for a specific time period or verify whether a specific data hash has been uploaded to the blockchain and its location on the blockchain. The blockchain trusted storage module outputs a verified time series data set; the on-chain verification records corresponding to the time series data set can be queried and verified for data integrity by querying the data function queryData (timestamp) → data hash value.

[0025] The model prediction module predicts future environmental and market factors based on raw data, providing a decision basis for dynamic power allocation; In one embodiment of the present invention, the processing of the model prediction module includes the following steps: S301, federated learning training; The model training unit uses the federated learning framework to train the prediction model; Federated learning allows multiple nodes to collaboratively train models without sharing original data, protecting data privacy. Each edge computing node in the system, as a participant in federated learning, trains a local model based on local data and then uploads only the model parameters (not the original data) to a central server for aggregation. The initial input of the model is multidimensional time series data, including characteristics such as photovoltaic power generation, wave height, wave period, sunshine intensity, seawater salinity, load demand, electricity price and carbon price. These data are normalized and then input into the network in chronological order. The model structure uses a deep learning architecture that combines LSTM (Long Short-Term Memory) with a recall mechanism. The LSTM layer is responsible for capturing the long-term dependencies of time series data. Its core consists of an input gate, a forget gate, and an output gate. The specific working principle is as follows: The forget gate determines which information needs to be discarded from the previous state. It connects the hidden state of the previous moment and the current input, processes it through the sigmoid function, and outputs a value between 0 and 1 to control the degree of retention of the cell state at the previous moment. The input gate determines what information needs to be updated; it also uses the hidden state of the previous moment and the current input to generate a control value through the sigmoid function and create a new candidate cell state (processed by the tanh function); The system updates the cell state by multiplying the previous state with the output of the forget gate (discarding some information) and adding the new candidate state controlled by the input gate (adding new information). The output gate controls what information is output based on the updated cell state; it uses the sigmoid function to process the combination of the previous hidden state and the current input, and then multiplies it with the cell state processed by the tanh function to generate the hidden state output at the current moment; The attention mechanism is used to highlight important time steps, and its workflow is as follows: The system calculates the energy score at each time step by transforming the current hidden state and the previous decoder state through a weight matrix, applying the tanh function, and then multiplying it by the learnable parameter vector; Use the softmax function to convert these energy scores into attention weights, ensuring that the sum of all weights is 1; The system uses these attention weights to perform a weighted summation of all hidden states to generate a context vector for subsequent predictions. The training process of federated learning uses the FedAvg algorithm, which implements distributed model training by training the model on multiple local nodes and aggregating parameter updates; The system begins with a central server initializing global model parameters, which are then distributed to participating nodes. Each node uses local data to train the model and calculate parameter updates, which are then sent back to the central server. The central server then takes a weighted average of all parameter updates based on the amount of data from each node to generate a new global model. This process repeats until the model converges or a predetermined number of training rounds are reached. S302, model compression and deployment; The trained model is compressed and deployed to edge computing devices. Model compression uses two main techniques: network pruning and parameter quantization. Network pruning reduces the model size by removing connections or neurons that contribute less to the prediction results. The system calculates the importance score of each connection, removes connections with scores below a threshold, and then fine-tunes the pruned network to restore model performance. Parameter quantization converts 32-bit floating-point parameters into 8-bit fixed-point representation, reducing model storage space and computational complexity. The system first determines the maximum absolute value of each layer parameter, calculates the scaling factor, and then quantizes the parameters to 8-bit integers. During inference, the parameter approximation is restored through dequantization operations. By using the above compression technology, the system reduces the model size by about 3 / 4, significantly reducing inference time and meeting the resource constraints of edge computing devices. S303, online prediction; The compressed and deployed model is called regularly for prediction. The system triggers a prediction task every hour, inputs the cleaned raw data of the past 168 hours (7 days), and generates a prediction sequence for the next 24 hours.

[0026] The prediction process normalizes the input data and maps each eigenvalue to the interval [-1, 1]; The system organizes the normalized input data into sliding windows with a window size of 24 (representing 1 day) and a step size of 1. For each window, the system calculates the model output through forward propagation to obtain the predicted value for the next 24 hours. Finally, the system denormalizes the model output to restore it to the original data scale; S304, interpretable output; The system uses the Shape Additive Explanations method (SHAP, SHapley Additive exPlanations) to analyze the factor contributions of each prediction value. SHAP is based on the concept of Shapley value in game theory and calculates the marginal contribution of each feature to the model prediction. The system generates a SHAP explanation for each predicted value, including a ranking of the contributions of various factors (such as sunshine intensity and wave height) and a visual chart to help users understand the model's decision-making process. Because the calculation of shape additive explanations involves complex combinatorial mathematics, the system uses an approximate method of the deep shape additive explanation algorithm. The specific steps are as follows: The system treats the deep network as a multi-layer composite function, which is calculated sequentially from the input layer to the output layer; For a given predicted value , system calculation characteristics The shape additive explanation value of : ; in, represents the shape additive explanation value of feature i; represents the set of all features, Indicates that features are not included A subset of represents factorial; represents the predicted value of the subset containing feature i; represents the predicted value of the subset that does not contain feature i; In one embodiment of the present invention, since the computational complexity of directly calculating the above formula is too high, the system adopts a recursive method, starting from the output layer, and backpropagating the contribution layer by layer; For each layer, the system calculates the contribution of each neuron in that layer to the next layer and distributes the contribution to the neurons in the previous layer; Through multiple recursive calculations, the system finally obtains the contribution of each feature of the input layer to the prediction result, that is, the shape additive explanation value; The system ranks each input feature (such as sunshine intensity, wave height, wave period, seawater salinity, etc.) according to its contribution and generates visual representations such as horizontal bar charts.

[0027] This recursive calculation method reduces computational complexity and enables the system to generate interpretable results in real time on edge computing devices, helping users understand the model's prediction basis under different environmental conditions and improving system transparency and credibility.

[0028] S305, adaptive fine-tuning; Monitor the model's prediction performance and update it when necessary (when the mean absolute percentage error between the predicted and observed values ​​exceeds a preset threshold, such as 8%). The system periodically (e.g., daily) compares the predicted values ​​with the observed values ​​and calculates the mean absolute percentage error (MAPE): ; in, represents the mean absolute percentage error, represents the actual value of the i-th sample, represents the predicted value of the i-th sample, Indicates the sample size.

[0029] When the MAPE exceeds a preset threshold (e.g., 8%), the system triggers model fine-tuning; The fine-tuning process uses the raw data from the last 24 hours to perform small batch gradient descent and update the model parameters; Fine-tuning is performed for only a small number of iterations (e.g., 5-10 rounds) to balance the update speed and computing resource consumption; After fine-tuning, the system re-evaluates model performance; if performance improves, the updated model is retained; otherwise, the system reverts to the original model and issues an alert, indicating that manual intervention or complete retraining may be required. The output of the model prediction module is 24-hour forecast data, which includes the forecast data for the next 24 hours (wave height time series, irradiance time series, salinity time series, electricity price time series, carbon price time series) and corresponding explanatory information.

[0030] Dynamic power allocation module, used to dynamically optimize system power allocation based on model prediction results to maximize carbon reduction benefits; In one embodiment of the present invention, the processing of the dynamic power allocation module includes the following steps: S401: Initial solution generation; Quickly generate feasible initial scheduling solutions based on simple rules; the system uses a rule engine to process input data and generate initial solutions based on the following heuristic rules: Photovoltaic power generation priority rule: during periods of high sunshine intensity, photovoltaic power generation power is allocated to the maximum value first; Desalination load matching rules: desalination power should match water supply demand as much as possible to ensure water supply; Electricity price response rules: During peak electricity price periods, desalination power is appropriately reduced to reduce electricity costs; Carbon price response rules: when carbon prices are high, increase photovoltaic utilization and reduce carbon emissions; Safety margin rules: retain about 15% power margin in each period to cope with emergencies; Based on these rules, the system generates an initial 24-hour scheduling plan, including hourly PV power allocation and desalination power allocation. The initial solution generation process takes less than 0.5 seconds, ensuring that a feasible solution can be quickly generated even in extreme situations. S402, deep optimization; Using the initial solution as a seed, an iterative optimization algorithm is used to generate a better scheduling solution. The system provides two optimization algorithms: an improved particle swarm optimization algorithm and an actor-critic reinforcement learning algorithm. The choice of algorithm is based on the specific scenario. For the improved particle swarm optimization algorithm, the system represents the scheduling plan as a 48-dimensional vector (24 hours × 2 control variables). The particle position represents the power allocation plan, and the particle velocity represents the update direction of the plan. The algorithm uses adaptive inertia weights and hybrid acceleration coefficients to enhance global search capabilities. The optimization objective function is: ; in, represents the optimization objective function value, represents the time step (hours), represents the carbon cost weight, represents the economic benefit weight, represents the carbon emission reduction per unit water production at time t, represents the carbon emission reduction at time t, represents the water production at time t, represents the economic benefits at time t; The particle swarm optimization algorithm searches for the optimal solution by iteratively updating the particle position and velocity. In each iteration, the system evaluates the fitness of each particle (the objective function value), updates the individual optimal position and the global optimal position, and then adjusts the particle velocity and position according to the update formula. The system also checks the constraints and corrects particles that violate them. The iterative process continues until the maximum number of iterations is reached or the convergence condition is met, and finally outputs the global optimal position as the optimized scheduling solution. Optionally, for the actor-critic reinforcement learning algorithm, the system models the scheduling problem as a Markov decision process; the state space contains information such as the current time, environmental predictions, and system parameters; the action space is the distribution of photovoltaic power and desalination power; and the reward function is the optimization objective function. The system uses two neural networks: an actor network that selects actions based on states, and a critic network that evaluates the value of state-action pairs. Through temporal difference learning and policy gradient methods, the system continuously optimizes the scheduling strategy and ultimately generates high-quality scheduling solutions. In one embodiment of the present invention, no matter which optimization algorithm is selected, the system will perform constraint checks on the generated scheduling plan to ensure that the photovoltaic power and desalination power do not exceed the maximum capacity and that the water production meets the basic requirements; For example, in a typical sunny summer day scheduling scheme, the system allocates PV power to near-maximum capacity (approximately 90%) between 9:00 AM and 3:00 PM, fully utilizing sunlight resources. Simultaneously, based on the water demand curve, the system increases desalination power to over 80% during peak water use periods (7:00 AM to 9:00 AM and 6:00 PM to 8:00 PM) to ensure adequate water supply. During the nighttime off-peak period (11:00 PM to 5:00 AM), desalination power is reduced to a baseline operating level of approximately 30%, while retaining approximately 15% of system capacity as emergency standby. This dynamic scheduling scheme satisfies basic constraints while achieving efficient resource utilization and minimizing carbon emissions.

[0031] S403, digital twin verification; The optimized candidate solutions are imported into a digital twin model for simulation verification. A digital twin model is a high-precision software replica of the physical system, incorporating detailed device characteristics, system dynamics, and environmental influencing factors.

[0032] The system first applies the candidate scheduling scheme to the digital twin model and simulates the system operation for 1 hour; During the simulation, the system records the actual response curves of key parameters, including actual photovoltaic output power, actual power consumption of the desalination system, water production, system losses, etc. Next, the system calculates the deviation between the simulation results and the expectations: ; ; ; in, represents the relative deviation of photovoltaic power at time t, represents the simulated photovoltaic power at time t, represents the planned PV power at time t; represents the relative deviation of the fade power at time t, represents the simulated fade power at time t, represents the planned fade power at time t; Indicates the relative deviation of water production at time t, represents the simulated water production at time t, Indicates the planned water production at time t.

[0033] S404, failover; The system is equipped with fault detection and emergency response mechanisms to ensure that basic operations can be maintained in the event of equipment abnormalities or communication interruptions; Real-time monitoring of device status signals and communication link quality. When an anomaly is detected (such as device alarm, sensor failure, communication timeout, etc.), it automatically switches to a predefined "fixed mode" backup strategy. Fixed-mode strategies are pre-set based on historical operating data and expert experience, and include several contingency plans for typical scenarios, such as "maximum safety mode" (all equipment operates at safe parameters), "load priority mode" (prioritizing water demand), and "generation priority mode" (prioritizing maximizing photovoltaic utilization). The system selects the appropriate backup strategy based on the type and severity of the fault to ensure safe and stable system operation. When the system returns to normal, it will automatically switch back to the optimized scheduling mode and record the fault event and handling process for subsequent analysis and improvement; The dynamic power allocation module outputs an optimized 24-hour schedule. This optimized and verified 24-hour schedule includes information such as (optimized PV power time series, optimized desalination power time series, water production time series, and reserved capacity), and is sent to the control system as an execution instruction. The carbon emission reduction benefit assessment module is used to calculate the carbon emission reduction benefit indicators of low-carbon projects and provide a quantitative basis for investment decisions; In one embodiment of the present invention, the processing of the carbon emission reduction benefit assessment module includes the following steps: S501, calculation of daily cumulative carbon reduction; The carbon reduction calculation unit calculates the system's daily cumulative carbon reduction based on the optimized scheduling plan. The carbon reduction calculation is based on two parts: Carbon emission reduction brought about by photovoltaic power generation replacing traditional energy and carbon emissions in the seawater desalination process; the calculation formula is: ; in, represents the daily cumulative carbon reduction, represents the optimized photovoltaic power at time t, The carbon emission reduction factor of photovoltaic power generation is the amount of carbon emissions reduced per kilowatt-hour of photovoltaic power generation compared with traditional energy; represents the water production at time t, The carbon emission factor for seawater desalination is the amount of carbon emissions per cubic meter of desalinated seawater; The carbon factor is determined using a life cycle assessment method, which comprehensively considers carbon emissions throughout the entire process of equipment manufacturing, transportation, installation, operation, and decommissioning. The system supports dynamic adjustment of the carbon factor value based on factors such as equipment batch, operating status, and environmental conditions to improve calculation accuracy. S502, calculation of total investment cost; Calculate the total investment cost of the system; the investment cost includes equipment purchase cost, installation and commissioning cost, operation and maintenance cost, etc. The calculation formula is: ; in, represents the total investment cost, Indicates the maximum capacity of the photovoltaic system, represents the unit photovoltaic capacity investment cost; Indicates the maximum capacity of the desalination system, represents the unit desalination capacity investment cost; represents the regional adjustment coefficient, which reflects the cost differences in different regions; m represents the operation and maintenance coefficient, which represents the annual operation and maintenance cost per unit capacity.

[0034] The system supports adjusting various parameters based on the actual project situation, taking into account the impact of factors such as equipment depreciation, technological progress, and subsidy policies to ensure the accuracy of investment cost calculations. At the same time, the system maintains a dynamically updated cost database that records the latest market prices of equipment from different suppliers and specifications, providing a reference for investment cost estimation. S503, generating investment intensity index; The carbon reduction investment intensity index is calculated based on the daily cumulative carbon reduction amount and the total investment cost. The calculation formula for investment intensity is: ; in, represents the daily cumulative carbon reduction, 365 represents the number of days in a year, represents the total investment cost, Indicates the intensity of carbon reduction investment, measured in tons of carbon dioxide per 10,000 yuan per year. This indicator indicates the amount of carbon emissions reduction that can be achieved per 10,000 yuan of investment per year. A higher value indicates a better emission reduction benefit of the investment. The system supports the calculation of investment intensity indicators based on different time scales, including daily, weekly, monthly and annual indicators, reflecting the changes in investment benefits over different time periods. At the same time, the system provides auxiliary financial indicators such as payback period and internal rate of return to help decision makers comprehensively evaluate the economic and environmental benefits of projects. The carbon emission reduction benefit assessment module outputs a daily report, which includes information such as (date, daily carbon reduction, total investment, and investment intensity).

[0035] Intelligent early warning and decision support module, used to monitor system operation status, identify potential problems and provide decision-making suggestions; In one embodiment of the present invention, the processing process of the intelligent early warning and decision support module includes the following steps: S601, threshold monitoring; Regularly scan system operation data to identify anomalies. The system performs a scan every 15 minutes, checking the following key indicators: Carbon reduction difference, calculate the difference between the actual carbon reduction amount and the planned value ,when When triggering an early warning; The calculation formula is: ; in, represents the carbon reduction difference, Indicates the actual carbon reduction amount, Indicates the planned carbon reduction amount; The investment intensity is abnormal. Check whether the real-time calculated investment intensity ρ exceeds the preset upper limit. ,when When triggering an early warning; Price fluctuations: Check the fluctuations of market prices (electricity prices, carbon prices). When the price changes exceed the standard deviation The system calculates the standard deviation based on historical price data. ; The specific method is to first calculate the average price data of the past 30 days, then calculate the square of the difference between each price point and the average, sum these squared differences and divide them by the number of samples, and finally take the square root to get the standard deviation; By default, the system uses the past 30 days' price data to calculate the standard deviation, which is updated daily. When the absolute difference between the real-time price and the price of the previous day at the same time exceeds twice the standard deviation (this multiple is a configurable sensitivity coefficient, with a default value of 2), the system triggers a price fluctuation alert. The system uses a sliding time window for statistical analysis to reduce the impact of short-term fluctuations. At the same time, the system supports dynamic threshold adjustment, automatically adjusting the warning threshold according to factors such as season and weather to improve the accuracy of warnings. S602: Optimization suggestions; After an anomaly is detected, the dynamic power allocation module is called to regenerate the scheduling plan; the system adopts different optimization strategies according to the type of anomaly: Abnormal carbon reduction differences: When the actual carbon reduction amount is significantly lower than the planned value, the system increases the α weight (carbon cost weight) and reduces the β weight (economic benefit weight), re-optimizes the scheduling plan, and improves the carbon reduction benefit; Abnormal investment intensity: When the investment intensity exceeds the preset upper limit, the system adjusts the optimization objective function, adds investment return constraints, and generates a more economical and efficient scheduling plan; Abnormal price fluctuations. When market prices change significantly, the system re-forecasts future trends based on the latest price data and adjusts optimization parameters to adapt to market changes. The system generates optimization recommendations that include new scheduling plans, the specific reasons for the changes, and the expected effects. These recommendations include both short-term countermeasures (such as power adjustments for the remainder of the day) and medium- and long-term optimization strategies (such as equipment maintenance recommendations and system upgrade options). S703, notification push; Push warning information and optimization suggestions to relevant personnel through multiple channels; the system supports multiple notification methods such as SMS, email, and mobile push, and the appropriate push method is selected according to the warning level and user role; The push content includes {warning type, warning time, original scheduling plan, optimized scheduling plan} and other information. Warning types include carbon reduction differences, abnormal investment intensity or price fluctuations; The system supports a tiered push strategy, determining push recipients and methods based on the urgency and impact of alerts. For example, general alerts are pushed to operations personnel via the app; important alerts are sent to department heads via both SMS and email; and emergency alerts are pushed to all relevant personnel through all channels, triggering audible and visual alarms.

[0036] The system also implements an intelligent anti-fatigue mechanism to avoid repeated push of similar warnings in a short period of time, reducing the burden on users; When similar alerts are detected multiple times within a short period of time, the system will merge these alerts into a single comprehensive message and reduce the push frequency appropriately. The output of the intelligent early warning and decision support module is an early warning log, which includes information such as the warning timestamp, warning type, original investment intensity, and the new 24-hour scheduling plan.

[0037] In one embodiment of the present invention, an application example of a typical low-carbon project carbon reduction investment intensity calculation system is provided: To verify the technical effectiveness of this invention, the proposed carbon reduction investment intensity calculation system was applied to a combined offshore photovoltaic and seawater desalination project. The project includes a 5MW offshore photovoltaic power generation system and a 10,000-ton-per-day seawater desalination system, deployed in the waters near an island in the East China Sea. The application test was conducted in a real production environment over a three-month period, covering various weather conditions and load scenarios. During the test, the system collected real-time data such as photovoltaic output, wave parameters, and water production, while also recording fluctuations in electricity and carbon prices. The test environment was equipped with a complete sensor network, including photovoltaic power plant energy meters, wave observation buoys, radiometers, seawater conductivity meters, load monitoring devices, and electricity / carbon price data interfaces. The data collection frequency was set to every five minutes, and over 100,000 raw data records were collected during system operation. The system first cleans and preprocesses the collected data, identifying and correcting approximately 3% of abnormal data points. The processed data is stored on the blockchain system, ensuring data credibility. The model prediction module trains an LSTM+Attention model based on the raw data to predict environmental and market factors for the next 24 hours, with prediction accuracy ranging from 92% to 95% under different weather conditions. Table 1 shows the prediction accuracy of the AI ​​prediction module under different weather conditions. Table 1: AI prediction accuracy statistics under different weather conditions The dynamic power allocation module uses an improved PSO algorithm to optimize system operating parameters, maximizing carbon reduction while meeting water supply demand. The system-generated scheduling plan is verified and executed using a digital twin, with actual operational deviations controlled within ±3%. By comparing traditional fixed scheduling plans with the system's dynamic optimization scheduling solution, key metrics such as photovoltaic utilization and system carbon reduction were collected, demonstrating the significant advantages of this system solution.

[0038] As shown in Table 2, the performance comparison between traditional fixed scheduling and dynamic optimization scheduling is shown; Table 2: Performance comparison between traditional fixed scheduling and dynamic optimization scheduling The carbon reduction benefit assessment module calculated an investment intensity index of 1.65tCO2 / 10,000 yuan / year, clearly quantifying the relationship between the project's environmental and economic benefits. The system also recorded monthly changes in investment intensity during the test period, reflecting the impact of seasonal changes on carbon reduction benefits. As shown in Table 3, the monthly trend of carbon reduction investment intensity is shown; Table 3: Monthly trends in carbon reduction investment intensity As shown in Table 4, the relationship between sunshine intensity and photovoltaic power generation is shown. The analysis results show that the dynamic power allocation strategy of this system can more effectively utilize variable sunshine resources; Table 4: Analysis of the relationship between sunshine intensity and photovoltaic power generation As shown in Table 5, the relationship analysis between the energy consumption and water production of the desalination system is presented, which also shows that the optimized scheduling of this system can reduce energy consumption while ensuring water supply demand; Table 5: Desalination system energy consumption and water production The test results show that compared with the traditional fixed scheduling strategy, the dynamically optimized scheduling scheme of this system significantly improves the photovoltaic utilization rate and the overall efficiency of the system; under typical sea conditions, the system's carbon reduction is significantly improved compared with the traditional scheme; at the same time, the water production cost is effectively reduced, and the system operation stability is also significantly improved; the carbon reduction investment intensity index clearly quantifies the relationship between the environmental and economic benefits of the project, providing a reliable basis for subsequent investment decisions.

[0039] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of equivalent embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. Typical low-carbon project carbon reduction investment intensity calculation system, characterized by: include: Data acquisition and preprocessing module, used to collect raw data from low-carbon project sites and perform preprocessing; Blockchain trusted storage module, used to ensure the integrity and traceability of original data and prevent data tampering; The model prediction module predicts future environmental and market factors based on raw data, providing a decision basis for dynamic power allocation; Dynamic power allocation module, used to dynamically optimize system power allocation based on model prediction results to maximize carbon reduction benefits; The carbon emission reduction benefit assessment module is used to calculate the ratio of carbon emission reduction to investment cost of low-carbon projects in real time and quantify the carbon reduction investment intensity index; Intelligent early warning and decision support module, used to monitor system operation status, identify potential problems and provide decision-making suggestions.

2. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: The data acquisition and preprocessing modules include: Multi-sensor redundant data collection, including active and standby sensor pairs, is used to alternately collect information on photovoltaic output, wave height, wave period, sunshine intensity, seawater salinity, campus load, and electricity and carbon prices; Multi-link transmission fusion, including submarine cable, satellite, and 4G / 5G transmission links, automatically selects the optimal transmission path through a weight function; Time synchronization, using GPS and PTP dual time calibration mechanism to ensure the time consistency of each collection point; Unified format: data from different sources and formats are converted into standard JSON format through a protocol gateway; Missing and abnormal data processing uses sliding window statistics combined with machine learning to fill in missing data and correct abnormal data.

3. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: The blockchain trusted storage module includes: Data hashing: Use the SHA-256 algorithm to calculate the hash value of the time series data set and call the smart contract to upload the hash value to the blockchain network; Consensus nodes use a consortium chain architecture and select PBFT or Raft consensus algorithms based on the characteristics of the deployment environment to ensure that all nodes reach consensus on the data hash records; The on-chain index uses a B+ tree structure to organize the timestamp index, maintains a mapping table from hash value to transaction ID, and supports efficient time range queries and hash value queries.

4. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: The model prediction module includes: Federated learning training uses the FedAvg algorithm to collaboratively train the prediction model on multiple nodes. The prediction model structure adopts a deep learning architecture that combines LSTM with an attention mechanism. Model compression deployment: deploying the model to edge computing devices through network pruning and parameter quantization; Online prediction: trigger a prediction task every hour, input the cleaned raw data of the past 168 hours, and generate a prediction sequence for the next 24 hours; Interpretable output, using the shape additive interpretation method to analyze the factor contribution of each predicted value; Adaptive fine-tuning triggers model fine-tuning based on the mean absolute percentage error (MAPE).

5. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: The dynamic power allocation module includes: Initial solution generation: quickly generate feasible initial dispatch plans based on PV generation priority, desalination load matching, electricity price response, carbon price response, and safety margin rules; Deep optimization uses an improved particle swarm optimization algorithm or actor-critic reinforcement learning algorithm to generate a better scheduling solution. The optimization objective function is: ； in, represents the optimization objective function value, represents the time step, represents the carbon cost weight, represents the economic benefit weight, represents the carbon emission reduction per unit water production at time t, represents the carbon emission reduction at time t, represents the water production at time t, represents the economic benefits at time t; Digital twin verification: importing the optimized candidate solutions into the digital twin model for simulation verification; Failover: Automatically switch to a predefined fixed-mode backup strategy in the event of device anomalies or communication interruptions.

6. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: The carbon reduction benefit assessment module includes: Carbon emission reduction calculation is used to calculate the daily cumulative carbon reduction of the system. The calculation formula is: ； in, represents the daily cumulative carbon reduction, represents the optimized photovoltaic power at time t, represents the carbon emission reduction factor of photovoltaic power generation; represents the water production at time t, represents the carbon emission factor for seawater desalination; Investment cost calculation, calculate the total investment cost of the system, including equipment purchase cost, installation and commissioning cost, and operation and maintenance cost; The investment intensity index is generated and the carbon reduction investment intensity index is calculated based on the daily cumulative carbon reduction amount and the total investment cost.

7. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 6 is characterized in that: The formula for carbon reduction investment intensity calculated by the investment intensity index generation unit is: ； in, represents the daily cumulative carbon reduction, represents the total investment cost, It represents the intensity of carbon reduction investment, with the unit being tons of carbon dioxide per 10,000 yuan per year.

8. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 1 is characterized in that: Intelligent early warning and decision support modules include: Threshold monitoring: regularly scanning system operation data to identify carbon reduction discrepancies, abnormal investment intensity, and abnormal price fluctuations; Optimization suggestion: call the dynamic power allocation module to regenerate the scheduling plan according to the exception type; Notification push: push warning information and optimization suggestions to relevant personnel via SMS, email, and mobile push channels.

9. The typical low-carbon project carbon reduction investment intensity calculation system according to claim 8 is characterized in that: The carbon reduction difference calculated by the threshold monitoring unit is: ； in, represents the carbon reduction difference, Indicates the actual carbon reduction amount, Indicates the planned carbon reduction amount.

10. A computer-readable storage medium, characterized in that It is used to store computer-readable instructions, and when the computer-readable instructions are read by a computer, it can run the typical low-carbon project carbon reduction investment intensity calculation system as described in any one of claims 1-9.

Images