Method and system for predicting investment and payback period of refrigerating machine room

Abstract

The invention discloses a refrigerating machine room investment and payback period prediction method and system, and belongs to the technical field of refrigerating machine room investment evaluation. The method solves the problem that the prior art depends on static empirical estimation and cannot give consideration to the influence of operating characteristics, energy consumption change and external environment factors of the refrigerating machine room on the investment payoff period, establishes a multi-dimensional mapping data table, associates equipment, operation, environment and market parameters with economic benefits, and introduces a virtual sample generation mechanism, so that the investment payoff period of the refrigerating machine room is optimized. A virtual and reasonably existing scene is extrapolated through a physical rule, so that virtual features and samples are generated, and the coverage range and diversity of a mapping data table are expanded; the influence of key parameters on the investment payback period is analyzed and evaluated through multi-dimensional sensitivity, so that investors can formulate risk response measures in advance under different scenes, the investment strategy is optimized, and the anti-risk capability and the investment return rate of a project are improved.

Description

Technical Field

[0001] This invention relates to the field of investment assessment technology for refrigeration plant, specifically a method and system for predicting investment and payback period for refrigeration plant. Background Technology

[0002] As an indispensable component of large buildings or industrial facilities, refrigeration rooms have high investment costs, and the length of the investment recovery period directly affects the economic benefits of enterprises.

[0003] Traditional investment assessment methods often rely on static experience estimations, which cannot take into account the operating characteristics of the chiller room, changes in energy consumption, and the impact of external environmental factors on the investment payback period. As a result, it is difficult to accurately predict the investment payback period of the chiller room, leading to significant risks in investment decisions.

[0004] Therefore, not content with existing needs, this paper proposes a method and system for predicting investment and payback period for chiller rooms. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for predicting investment and payback period for refrigeration plant facilities. By establishing a multi-dimensional mapping data table, it correlates equipment, operation, environment, and market parameters with economic benefits. A virtual sample generation mechanism is introduced to extrapolate virtual but plausible scenarios from physical rules, thereby generating virtual features and samples, expanding the coverage and diversity of the mapping data table. Through multi-dimensional sensitivity analysis, the impact of key parameters on the investment payback period is assessed, enabling investors to anticipate different scenarios and formulate risk response measures in advance, optimize investment strategies, improve the project's risk resistance and return on investment, and solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for predicting investment and payback period for refrigeration plant facilities includes the following steps: Collect equipment and operating parameters of the refrigeration room from historical data, as well as market cost parameters and environmental climate parameters, as a historical multi-source dataset; Historical multi-source datasets were organized and a mapping data table was established to reflect the annual net income and investment achieved by the refrigeration room under different operating parameters, environmental climate parameters and market cost parameters. Convolutional neural network model is constructed, with equipment parameters, operating parameters, environmental and climate parameters, and market cost parameters as input features, and the annual net income and investment payback period of the refrigeration room as output targets; The neural network model is trained and tested using historical multi-source datasets. A time attention mechanism is introduced to automatically identify and weight the factors affecting energy consumption, such as energy price fluctuations and equipment aging, and dynamically adjust the input features to calculate the annual net income of the chiller room. The investment payback period of the refrigeration plant can be predicted based on the ratio of its annual net revenue to its total investment cost.

[0007] Further, a mapping data table is established, including: Preprocessing of historical multi-source datasets includes: data cleaning and time scale alignment; Based on correlation analysis, features related to the prediction of investment and payback period of refrigeration room are extracted from historical multi-source datasets, including: equipment static features, operating mode features, environmental and climatic features, and market cost features. The static characteristics, operational modal characteristics, environmental and climatic characteristics, and market cost characteristics of the equipment are concatenated to obtain the state vector; Based on the static and operational characteristics of the equipment in the chiller room, combined with environmental and climatic characteristics and market cost characteristics, an energy consumption model for the chiller room is established, and real-time energy consumption is calculated. Based on the real-time energy consumption corresponding to the state vector, set the corresponding economic benefit label.

[0008] Furthermore, based on the real-time energy consumption corresponding to the state vector, after setting the corresponding economic benefit labels, including: Sampling is performed on the state vectors of historical multi-source datasets. Based on their economic benefit labels, extrapolation scenarios are pre-defined, and virtual features that do not appear in historical multi-source datasets but still exist reasonably are generated. The virtual features are input into the energy consumption model to calculate its virtual energy consumption, and corresponding virtual economic benefit labels are set. Virtual features and virtual economic benefit labels are used as virtual samples and injected into the mapping data table to expand the coverage and diversity of the mapping data table; Virtual features are fused with state vectors to form optimized feature vectors.

[0009] Furthermore, a time attention mechanism is introduced to automatically identify and weight the impact factors of energy price fluctuations and equipment aging on energy consumption, dynamically adjusting input features, including: Energy price fluctuations and equipment aging in the optimized feature vector are used as time series features. Time series analysis methods are used to extract time series features so that the feature vector reflects the dynamic changes of the time series. Through linear transformation, corresponding query vector, key vector, and value vector are generated. The query vector is the input feature vector, and the key vector and value vector are time series feature vectors. Attention weights are obtained by calculating the similarity between the query vector and the key vector. Weighted summation of the value vectors is then performed based on the attention weights to obtain the weighted time series features. The weighted time series features are concatenated with the input features in the optimized feature vector to form a new input feature vector.

[0010] Furthermore, after obtaining the weighted time series features, they include: In convolutional neural network models, new input feature vectors are dynamically updated; Continue to monitor energy price fluctuations and equipment aging, and confirm that new input characteristics can reflect the latest operating environment; The model's input features are adjusted based on real-time data, and the model's prediction results are dynamically updated.

[0011] Furthermore, based on the ratio of the annual net income of the chiller room to the total investment cost, the investment payback period of the chiller room is predicted, including: Calculate the total investment cost of the refrigeration room based on the equipment procurement cost, installation and commissioning cost, and operation and maintenance cost; Analyze the time value of money, discount future operation and maintenance costs, and calculate the present value of the total investment cost; Based on the annual net income of the refrigeration room predicted by the convolutional neural network model, the investment payback period of the refrigeration room is predicted.

[0012] Furthermore, after predicting the investment payback period for the chiller room, it also includes: Key parameters in the feature vectors are extracted for sensitivity analysis, and different assumptions are set for each parameter. For each assumption, recalculate the annual net income and payback period of the chiller room; Compare the changes in payback period under different assumptions and assess the sensitivity of each parameter to payback period; The results of the sensitivity analysis are compiled into a report, including the annual net income and investment payback period under each assumption, to provide a scientific basis for investment decisions.

[0013] Furthermore, a convolutional neural network model is constructed, including: Construct a convolutional neural network model, including multiple convolutional layers and pooling layers, to extract local and global features from the input features; The historical multi-source dataset is divided into training set, validation set and test set. The convolutional neural network model is trained using the training set. The model parameters are adjusted by optimization algorithm to minimize prediction error. The convolutional neural network model is validated using a validation set to evaluate its performance. Adjust the parameters of the convolutional neural network model and optimize its structure based on the validation results; Use the test set to test the trained model and evaluate its performance on unseen data.

[0014] A system for predicting investment and payback period for refrigeration plant rooms, comprising: The multi-source data acquisition and processing module is configured to collect equipment and operating parameters of the refrigeration room, as well as market cost parameters and environmental climate parameters, and perform data cleaning and time alignment processing. The feature extraction and mapping module is configured to extract equipment static features, operating mode features, environmental climate features and market cost features from multi-source data based on correlation analysis, concatenate them to form a state vector, and set corresponding economic benefit labels. The intelligent prediction module is configured to predict the state vector based on a convolutional neural network model and output the annual net income of the refrigeration room. Based on the ratio of the annual net income of the refrigeration room to the total investment cost, the investment payback period of the refrigeration room is predicted. The sensitivity analysis module is configured to perform sensitivity analysis on key parameters and recalculate the annual net income and investment payback period for each set assumption. It compares the changes in investment payback period under different scenarios and provides a scientific basis for investment decisions based on the results of the sensitivity analysis.

[0015] Furthermore, it also includes: The mapping extension module is configured to pre-define extrapolation scenarios based on economic benefit labels, generate virtual features and virtual economic benefit labels, use the virtual features and virtual economic benefit labels as virtual samples, and inject them into the mapping data table. The dynamic feature optimization module is configured to take energy price fluctuations and equipment aging as time series features, perform a weighted summation on them through linear transformation to obtain weighted time series features, and then concatenate the weighted time series features with the input features in the optimized feature vector to form a new input feature vector.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention improves the quality of feature vectors by establishing a multi-dimensional mapping data table that links equipment, operation, environment, and market parameters with economic benefits. It also introduces a virtual sample generation mechanism, extrapolating virtual but plausible scenarios from physical rules to generate virtual features and samples, expanding the coverage and diversity of the mapping data table and enhancing the model's generalization ability. Furthermore, it fuses virtual features with feature vectors, resulting in optimized feature vectors that support more robust and accurate investment return predictions, providing investors with a solid foundation.

[0017] 2. This invention assesses the impact of key parameters on the investment payback period through multi-dimensional sensitivity analysis, which can identify factors that have a significant impact on the project's economics, thereby accurately identifying major risk sources and optimization opportunities; and generates a sensitivity analysis report, enabling investors to formulate risk response measures in advance according to different scenarios, optimize investment strategies, improve the project's risk resistance and return on investment, and ensure that wise investment decisions are made in a complex market. Attached Figure Description

[0018] Figure 1 This is a flowchart of a method for predicting investment and payback period for a refrigeration room according to the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] To address the technical problem that traditional investment assessment methods often rely on static, empirical estimations, failing to consider the operational characteristics of chiller rooms, energy consumption variations, and the impact of external environmental factors on the payback period, thus making it difficult to accurately predict the payback period and leading to significant risks in investment decisions, please refer to [link to relevant documentation]. Figure 1 This embodiment provides the following technical solution: A method for predicting investment and payback period for refrigeration plant facilities includes the following steps: Collect historical data on equipment and operating parameters of refrigeration rooms, as well as market cost and environmental climate parameters, to create a historical multi-source dataset. Environmental climate parameters include building cooling load data and geographic climate data. Market cost parameters include market value, energy price data, target customers and pricing standards for refrigeration rooms, annual net income and investment payback period for refrigeration rooms. Equipment parameters include, but are not limited to, the model, power and cooling capacity of refrigeration units. Operating parameters include, but are not limited to, operating time, operating mode and equipment load rate.

[0021] Historical multi-source datasets were organized and mapped into tables reflecting the annual net income and investment of the chiller room under different operating parameters, environmental and climatic parameters, and market cost parameters, including: The historical multi-source dataset is preprocessed, including data cleaning and timescale alignment; ensuring accurate alignment of equipment operating parameters, environmental climate parameters, and market cost parameters in terms of timestamps; using correlation analysis, features relevant to the investment and payback period prediction of the chiller room are extracted from the historical multi-source dataset, including equipment static features, operating modal features, environmental climate features, and market cost features; these features are concatenated to obtain a state vector, providing comprehensive input data for subsequent modeling; based on the equipment static and operating modal features of the chiller room, combined with environmental climate and market cost features, an energy consumption model for the chiller room is established, and real-time energy consumption is calculated; for example, the energy consumption model is modified based on the impact of climate conditions on cooling demand and the impact of electricity price fluctuations on operating costs. Based on the real-time energy consumption corresponding to the state vector, appropriate economic benefit labels are set to provide target variables for subsequent prediction models.

[0022] Sampling is performed on the state vectors of historical multi-source datasets. Based on their economic benefit labels, extrapolation scenarios are pre-defined, such as: more energy-efficient equipment, more extreme climate parameters, and virtual features that do not appear in the historical multi-source datasets but still exist reasonably are generated. The virtual features are input into the energy consumption model to calculate its virtual energy consumption and set corresponding virtual economic benefit labels. The virtual features and virtual economic benefit labels are used as virtual samples and injected into the mapping data table to expand the coverage and diversity of the mapping data table, making the model trained later more generalizable. The virtual features are fused with the state vectors to form optimized feature vectors.

[0023] The beneficial effects achieved by the above are as follows: By establishing a multi-dimensional mapping data table, the parameters of equipment, operation, environment, and market are linked to economic benefits, thereby improving the quality of feature vectors; and by introducing a virtual sample generation mechanism, virtual but reasonably existing scenarios are extrapolated from physical rules, thereby generating virtual features and samples, expanding the coverage and diversity of the mapping data table, and enhancing the generalization ability of the model; and by fusing virtual features with feature vectors, the resulting optimized feature vectors can support the model in making more robust and accurate investment return predictions, providing investors with a solid basis.

[0024] A convolutional neural network model is constructed, using equipment parameters, operating parameters, environmental and climate parameters, and market cost parameters as input features, and the annual net income and investment payback period of the chiller room as output targets, including: A convolutional neural network (CNN) model is constructed, including multiple convolutional and pooling layers, to extract local and global features from the input features. Historical multi-source datasets are divided into training, validation, and test sets to ensure the independence and reliability of the training, validation, and testing processes. The CNN model is trained using the training set, and its parameters are adjusted through optimization algorithms to minimize prediction error and improve prediction accuracy. The CNN model is validated using the validation set to evaluate its performance and ensure its stability and reliability. Based on the validation results, the CNN model parameters are adjusted to optimize its structure and improve its generalization ability. The trained model is tested using the test set to evaluate its performance on unseen data, ensuring its prediction accuracy and reliability.

[0025] The neural network model was trained and tested using historical multi-source datasets, and a time attention mechanism was introduced to automatically identify and weight the impact factors of energy price fluctuations and equipment aging on energy consumption, dynamically adjusting the input features, including: Energy price fluctuations and equipment aging in the optimized feature vector are used as time series features. Time series analysis methods, such as the sliding window method, are used to extract these features, allowing the feature vector to reflect the dynamic changes of the time series. Through linear transformation, corresponding query vectors, key vectors, and value vectors are generated. The query vector serves as the input feature vector, while the key and value vectors are the time series feature vectors. Attention weights are obtained by calculating the similarity between the query vector and the key vector, reflecting the importance of each time point. The value vectors are then weighted and summed according to the attention weights to obtain weighted time series features, which better reflect the important information in the time series. Finally, the weighted time series features are concatenated with the input features in the optimized feature vector to form a new input feature vector.

[0026] In the convolutional neural network model, new input feature vectors are dynamically updated; energy price fluctuations and equipment aging are continuously monitored to confirm that the new input features can reflect the latest operating environment; the model's input features are adjusted according to real-time data to dynamically update the model's prediction results; and through an online learning mechanism, the model can dynamically adjust parameters based on new data to improve the model's adaptability and prediction accuracy.

[0027] Based on this, the annual net income of the refrigeration room is calculated; for example, the impact of changes in market demand on pricing standards and the impact of policy adjustments on income are considered to revise the income model; based on the ratio of the annual net income of the refrigeration room to the total investment cost, the investment payback period of the refrigeration room is predicted, including: The total investment cost of the chiller room is calculated based on the equipment procurement cost, installation and commissioning cost, and operation and maintenance cost. The present value of the total investment cost is calculated by analyzing the time value of money and discounting future operation and maintenance costs. The investment payback period of the chiller room is predicted based on the annual net income of the chiller room predicted by the convolutional neural network model.

[0028] Key parameters from the feature vector are extracted for sensitivity analysis, such as equipment aging, energy price fluctuations, changes in equipment efficiency, changes in market demand, and policy adjustments. Different assumptions are set for each parameter, such as a 10% increase in equipment aging, a 20% increase in energy prices, and a 15% increase in equipment efficiency. For each assumption, the annual net income and payback period of the chiller room are recalculated. For example, if equipment aging increases by 10%, energy costs and annual net income are recalculated. If energy prices increase by 20%, energy costs and annual net income are recalculated. If equipment efficiency increases by 15%, energy costs and annual net income are recalculated. The changes in payback period under different assumptions are compared to assess the sensitivity of each parameter to the payback period. For example, a 10% increase in equipment aging extends the payback period by 0.5 years. A 20% increase in energy prices extends the payback period by 1 year. A 15% increase in equipment efficiency shortens the payback period by 0.3 years. The results of the sensitivity analysis will be compiled into a report, including the annual net income and payback period under each assumption. Based on the results of the sensitivity analysis, a scientific basis for investment decisions will be provided. For example, if equipment aging has a significant impact on the payback period, regular equipment maintenance and replacement are recommended. If energy price fluctuations have a significant impact on the payback period, energy-saving measures or long-term energy contracts are recommended. If changes in market demand have a significant impact on the payback period, it is recommended to monitor market dynamics and adjust pricing and service models accordingly.

[0029] In one embodiment, suppose a company plans to invest in and construct a cooling room to provide cooling services for a large data center. The equipment selection for the cooling room includes two 1000kW cooling units, each with a power of 500kW. The cooling room operates continuously throughout the year, with load adjustments made according to the data center's business needs. The local climate is a temperate monsoon climate, and the electricity pricing policy is peak-valley pricing, with relatively small energy price fluctuations. The model predicts an annual net income of 15.02 million yuan and a payback period of 1.07 years. Based on this, key parameters are extracted, and the following assumptions are made: a 10% increase in equipment aging; a 20% increase in energy prices; and a 15% increase in equipment efficiency. Recalculations show that: a 10% increase in equipment aging extends the payback period by 0.5 years; a 20% increase in energy prices extends the payback period by 1 year; and a 15% increase in equipment efficiency shortens the payback period by 0.3 years. The resulting investment decision recommendation is: if equipment aging has a significant impact on the payback period, regular equipment maintenance and upgrades are recommended. If energy price fluctuations significantly impact the investment payback period, it is recommended to implement energy-saving measures or sign long-term energy contracts. If changes in market demand significantly impact the investment payback period, it is recommended to monitor market dynamics and adjust pricing and service models accordingly.

[0030] The beneficial effects achieved by the above are as follows: by assessing the impact of key parameters on the investment payback period through multi-dimensional sensitivity analysis, factors that have a significant impact on the project's economics can be identified, thereby accurately identifying major risk sources and optimization opportunities; and a sensitivity analysis report can be generated, enabling investors to formulate risk response measures in advance according to different scenarios, optimize investment strategies, improve the project's risk resistance and return on investment, and ensure that wise investment decisions are made in a complex market.

[0031] A system for predicting investment and payback period for refrigeration plant rooms, comprising: The multi-source data acquisition and processing module is configured to collect equipment and operating parameters of the refrigeration room, as well as market cost parameters and environmental climate parameters, and perform data cleaning and time alignment processing.

[0032] The feature extraction and mapping module is configured to extract equipment static features, operating mode features, environmental climate features, and market cost features from multi-source data based on correlation analysis, concatenate them to form a state vector, and set corresponding economic benefit labels.

[0033] The intelligent prediction module is configured to predict the state vector based on a convolutional neural network model and output the annual net income of the refrigeration room. Based on the ratio of the annual net income of the refrigeration room to the total investment cost, the investment payback period of the refrigeration room is predicted.

[0034] The sensitivity analysis module is configured to perform sensitivity analysis on key parameters and recalculate the annual net income and investment payback period for each set assumption. It compares the changes in investment payback period under different scenarios and provides a scientific basis for investment decisions based on the results of the sensitivity analysis.

[0035] The mapping extension module is configured to pre-define extrapolation scenarios based on economic benefit labels, generate virtual features and virtual economic benefit labels, use the virtual features and virtual economic benefit labels as virtual samples, and inject them into the mapping data table.

[0036] The dynamic feature optimization module is configured to take energy price fluctuations and equipment aging as time series features, perform a weighted summation on them through linear transformation to obtain weighted time series features, and then concatenate the weighted time series features with the input features in the optimized feature vector to form a new input feature vector.

[0037] Working Principle: By establishing a mapping data table, correlation analysis is used to extract features related to investment and payback period prediction, forming a state vector. An energy consumption model is then established to calculate real-time energy consumption and set economic benefit labels. Simultaneously, a convolutional neural network model is constructed, using equipment parameters, operating parameters, environmental and climate parameters, and market cost parameters as input features, and annual net income and payback period as output targets. Historical data is used for training, validation, and testing. A time attention mechanism is introduced to automatically identify and weight the impact of energy price fluctuations and equipment aging on energy consumption, dynamically adjusting the input features. The convolutional neural network model then predicts annual net income, and the payback period is calculated in conjunction with the total investment cost. Sensitivity analysis is performed to assess the impact of key parameters on the payback period, and the analysis results are compiled into a report, providing a scientific basis for investment decisions.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method of predicting the payback period of an investment in a chiller plant, characterized by, The method comprises the following steps: Collecting equipment parameters and operation parameters of the refrigeration machine room in historical data, as well as market cost parameters and environmental climate parameters, as a historical multi-source data set; Organizing the historical multi-source data set, and establishing a mapping data table reflecting the annual net income and investment of the refrigeration machine room under different operation parameters, environmental climate parameters and market cost parameters of the equipment; Building a convolutional neural network model, taking the equipment parameters, operation parameters, environmental climate parameters and market cost parameters as input features, and taking the annual net income and investment recovery period of the refrigeration machine room as output targets; Training and testing the neural network model using the historical multi-source data set, and introducing a time attention mechanism to automatically identify and weight the factors affecting energy consumption caused by energy price fluctuations and equipment aging, dynamically adjusting the input features, and calculating the annual net income of the refrigeration machine room accordingly; According to the ratio of the annual net income of the refrigeration machine room to the total investment cost, the investment recovery period of the refrigeration machine room is predicted.

2. The method for predicting investment and payback period of a chiller plant room according to claim 1, wherein, The mapping data table is established, including: Preprocessing the historical multi-source data set, including data cleaning and time scale alignment; According to the correlation analysis method, the features related to the refrigeration machine room investment and recovery period prediction are extracted from the historical multi-source data set, including equipment static features, operation modal features, environmental climate features and market cost features; Splicing the equipment static features, operation modal features, environmental climate features and market cost features to obtain a state vector; According to the equipment static features and operation modal features of the refrigeration machine room, combined with the environmental climate features and market cost features, an energy consumption model of the refrigeration machine room is established, and the real-time energy consumption is calculated; According to the real-time energy consumption corresponding to the state vector, the corresponding economic benefit label is set.

3. The method of predicting investment and payback period of a chiller plant room according to claim 2, wherein, After setting the corresponding economic benefit label according to the real-time energy consumption corresponding to the state vector, including: Sampling in the state vector of the historical multi-source data set, and according to its economic benefit label, an extrapolation scenario is preset, and a virtual feature that does not exist in the historical multi-source data set but still reasonably exists is generated; The virtual feature is input into the energy consumption model to calculate its virtual energy consumption, and the corresponding virtual economic benefit label is set; The virtual feature and the virtual economic benefit label are used as virtual samples, and are injected into the mapping data table to expand the coverage and diversity of the mapping data table; The virtual feature is fused with the state vector to form an optimized feature vector.

4. The method of predicting investment and payback period of a chiller plant room according to claim 3, wherein, The time attention mechanism is introduced to automatically identify and weight the factors affecting energy consumption caused by energy price fluctuations and equipment aging, dynamically adjusting the input features, including: Taking the energy price fluctuations and equipment aging in the optimized feature vector as time series features, the time series features are extracted through time series analysis method, so that the feature vector reflects the dynamic changes of the time series; Through linear transformation, the corresponding query vector, key vector and value vector are generated, the query vector is the input feature vector, and the key vector and value vector are the time series feature vector; The attention weight is obtained by calculating the similarity between the query vector and the key vector, and the weighted time series features are obtained by weighting and summing the value vector according to the attention weight; The weighted time series features are spliced with the input features in the optimized feature vector to form a new input feature vector.

5. The method of predicting the payback period of an investment in a refrigeration plant room according to claim 4, characterized in that, After obtaining the weighted time series features, the following steps are included: In the convolutional neural network model, the new input feature vector is dynamically updated; Continue to monitor the energy price fluctuations and the degree of equipment aging to confirm that the new input features can reflect the latest operating environment; Adjust the input features of the model according to the real-time data and dynamically update the prediction results of the model.

6. The method of predicting the payback period of an investment in a refrigeration plant room according to claim 5, characterized in that, According to the ratio of the annual net income of the refrigeration machine room to the total investment cost, the investment recovery period of the refrigeration machine room is predicted, including: According to the equipment procurement cost, installation and commissioning cost, and operation and maintenance cost, the total investment cost of the refrigeration machine room is calculated; Analyze the time value of money and discount the future operation and maintenance cost to calculate the present value of the total investment cost; According to the annual net income of the refrigeration machine room predicted by the convolutional neural network model, the investment recovery period of the refrigeration machine room is predicted.

7. The method of predicting the payback period of an investment in a refrigeration plant room according to claim 6, characterized in that, After predicting the investment recovery period of the refrigeration machine room, the following steps are included: Extract key parameters in the feature vector for sensitivity analysis and set different assumption conditions for each parameter; Recalculate the annual net income and investment recovery period of the refrigeration machine room for each assumption condition; Compare the changes in investment recovery period under different assumption conditions to evaluate the sensitivity of each parameter to investment recovery period; Organize the results of sensitivity analysis into a report, including the annual net income and investment recovery period under each assumption condition, and the results of sensitivity analysis.

8. The method for predicting investment and payback period of a chiller plant room according to claim 1, wherein, Constructing a convolutional neural network model includes: Constructing a convolutional neural network model includes multiple convolutional layers and pooling layers to extract local features and global features from input features; Divide the historical multi-source data set into training set, validation set and test set, use the training set to train the convolutional neural network model, adjust the model parameters through optimization algorithm to minimize the prediction error; Use the validation set to validate the convolutional neural network model and evaluate the performance of the convolutional neural network model; Adjust the convolutional neural network model parameters according to the validation results to optimize its structure; Test the trained model using the test set to evaluate the model's performance on unseen data.

9. A system for predicting investment and payback period of a chiller plant, applied to the method for predicting investment and payback period of a chiller plant according to any one of claims 1-8, characterized in that, It includes: A multi-source data acquisition and processing module is configured to collect equipment parameters and operating parameters of the refrigeration machine room, as well as market cost parameters and environmental and climate parameters, and to perform data cleaning and time alignment processing; A feature extraction and mapping module is configured to extract device static features, operating mode features, environmental and climate features, and market cost features from multi-source data based on correlation analysis, splice them to form a state vector, and set corresponding economic benefit labels; An intelligent prediction module is configured to predict the state vector based on a convolutional neural network model and output the annual net income of the refrigeration machine room, and predict the investment recovery period of the refrigeration machine room according to the ratio of the annual net income of the refrigeration machine room to the total investment cost; A sensitivity analysis module is configured to perform sensitivity analysis on key parameters and recalculate the annual net income and investment recovery period for each set assumption condition, compare the changes in investment recovery period under different scenarios, and analyze the results of sensitivity analysis.

10. The system for predicting investment and payback period of a chiller plant room according to claim 9, wherein, It also includes: The mapping expansion module is configured to, according to the economic benefit label, preset an extrapolation scenario, generate a virtual feature and a virtual economic benefit label, take the virtual feature and the virtual economic benefit label as a virtual sample, and inject the virtual sample into a mapping data table. The dynamic feature optimization module is configured to take the energy price fluctuation and the equipment aging as time sequence features, perform weighted summation on the time sequence features through linear transformation to obtain weighted time sequence features, splice the weighted time sequence features with input features in an optimization feature vector to form a new input feature vector.

Images