Multi-dimensional operation quantitative evaluation method for multi-season park comprehensive energy system

Abstract

The invention provides a multi-dimensional operation quantitative evaluation method for a multi-season park integrated energy system, and relates to the technical field of energy management and optimization, and the method comprises the steps: constructing a quantitative evaluation system, enabling the quantitative evaluation system to be coupled with ecological network analysis through a dissipation structure theory, and constructing a dynamic collaborative evaluation framework. In combination with social physics, time geography, thermodynamic constraint and cultural evolution theories, a six-dimensional evaluation system including self-organization characteristics, equipment collaboration, user group behavior influence, space-time coupling characteristics, theoretical performance limit and operation strategy adaptability is formed. In order to solve the problems that a traditional evaluation method is single in dimension and cannot comprehensively reflect system characteristics, a six-dimensional evaluation system of self-organization characteristics, equipment collaboration, user group behavior influence, space-time coupling characteristics, theoretical performance limit and operation strategy adaptability is constructed, and based on a dissipation structure theory, ecological network analysis and other advanced theories, the system characteristics are evaluated. And systematically evaluating the operation state of the integrated energy system.

Description

Technical Field

[0001] The present invention relates to the field of energy management and optimization technology, and specifically to a multi-dimensional operation quantitative evaluation method for a multi-season park integrated energy system. Background Art

[0002] With the rapid development of modern integrated energy systems, industrial parks, as important carriers integrating multiple energy forms such as heat, gas, and electricity, face significant challenges in terms of operational efficiency and multi-seasonal adaptability, which are directly related to the economic efficiency and sustainability of energy utilization and the stability of social operations. Multi-season industrial park integrated energy systems aim to achieve efficient energy scheduling and dynamic optimization by integrating multiple energy devices (such as generators, energy storage devices, and heating pipelines), user behavior data, and operational strategies. However, due to seasonal demand fluctuations (such as a surge in heating demand in winter and a dominant cooling demand in summer), the complexity of inter-device coordination, the diversity of user behavior, and the heterogeneity of temporal and spatial distribution, the system's operational characteristics vary significantly across seasons. Traditional evaluation methods struggle to fully capture these dynamic characteristics, resulting in a lack of scientific basis for optimization strategies. Therefore, developing a multidimensional, dynamic, and theory-driven quantitative evaluation method to systematically evaluate the performance of integrated energy systems in multi-seasonal operation has become a key research direction in the field of energy management.

[0003] Although existing energy system evaluation technologies have made some progress in park energy management, they still have the following shortcomings due to the complex characteristics of multi-season integrated energy systems:

[0004] Problem 1: The evaluation dimension is single and it is difficult to fully reflect the characteristics of the system; traditional evaluation methods usually focus on a single indicator, such as energy efficiency or economic cost, and ignore the multi-dimensional characteristics of the integrated energy system, including self-organization characteristics, equipment coordination, the influence of user group behavior, spatiotemporal coupling characteristics, theoretical performance limits and operational strategy adaptability. This single-dimensional analysis cannot capture the overall operating status of the system, especially under seasonal changes (such as thermal energy flow dominance in winter and electrical energy flow dominance in summer), the evaluation results lack comprehensiveness. For example, ignoring the impact of user behavior may cause the scheduling strategy to deviate from actual demand, resulting in more than 10% energy waste; ignoring equipment coordination may fail to identify key interaction nodes, reducing resource allocation efficiency by 15%.

[0005] The second problem is that the frequency of data collection is low, making it difficult to capture dynamic changes. Existing data collection methods mostly rely on low-frequency monitoring (such as daily or hourly collection), and are unable to obtain high-frequency energy flow data (such as changes in heat flow, gas flow, and electricity flow per second) or user behavior data (such as minute-by-minute electricity usage patterns) in real time. This results in the evaluation model's lack of accurate characterization of the system's dynamic behavior, making it difficult to provide real-time support, especially during seasonal fluctuations (such as a rapid increase in heating demand in winter) or emergencies (such as peak electricity consumption). For example, low-frequency collection may ignore fluctuations in heat flow within 1 hour, resulting in an entropy production rate calculation deviation of more than 5%, affecting the accuracy of the self-organizing characteristic evaluation.

[0006] The third problem is that theoretical models lack depth and multidisciplinary integration. Traditional evaluation methods are mostly based on simple statistical analysis or empirical formulas, and fail to fully integrate advanced theories such as dissipative structure theory, ecological network analysis, social physics, time geography, thermodynamic constraints, and cultural evolution theory. These theories are respectively applicable to analyzing a system's energy evolution, device interaction, user behavior, spatiotemporal distribution, performance limits, and strategic adaptability. A single model is unlikely to reveal the inherent laws of a system. For example, the lack of ecological network analysis may make it impossible to quantify the efficiency of energy interactions between devices, resulting in the lack of circulation rate assessment and affecting waste heat recovery optimization. The lack of a social physics model may ignore the consistency of user behavior and reduce the reliability of system responsiveness analysis.

[0007] Problem four: outdated data analysis methods lack accuracy and robustness. Existing technologies often rely on simple threshold judgments or static weight analysis, failing to fully utilize modern data analysis techniques such as ARIMA time series analysis, PageRank graph theory algorithms, TOPSIS multi-objective decision analysis, and the SIR propagation dynamic model. These techniques can accurately quantify indicators such as entropy production rate, reciprocity index, and comprehensive evaluation scores, while traditional methods are prone to misjudgment due to seasonal variations or data noise. For example, energy efficiency analysis based on fixed thresholds may ignore highly volatile winter data, resulting in a deviation of up to 20% in the comprehensive evaluation score. Furthermore, the lack of data cross-validation mechanisms (such as Pearson correlation analysis) makes it difficult to ensure consistency when integrating multi-source data, which may lead to an evaluation error of 10%.

[0008] Problem five: The evaluation results are presented in a single format, lacking intuitiveness and interactivity. Traditional evaluation reports are mostly output in the form of static tables or charts, which make it difficult to intuitively display the system's spatiotemporal distribution characteristics (such as spatial heat maps of resource allocation efficiency) or multi-seasonal trends (such as seasonal changes in entropy production). This makes it difficult for managers to quickly identify optimization potential. For example, inefficient areas may be overlooked due to the lack of spatial visualization, missing 15% of optimization opportunities. In addition, the report output format is single (such as PDF only) and lacks interactive functions (such as time sliders and indicator filtering). It cannot meet the diverse needs of different users (such as managers and engineers), limiting the practical application value of the evaluation results. Summary of the Invention

[0009] In response to the shortcomings of the existing technology, the present invention provides a multi-dimensional operation quantitative evaluation method for the multi-season park integrated energy system, which solves the problems in the existing technology of single evaluation dimensions and evaluation results, and unsatisfactory data collection and analysis and theoretical model effects.

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0011] The quantitative evaluation method for the multi-dimensional operation of the multi-season park integrated energy system includes the following steps:

[0012] Collect multi-source data, including heat flow data, gas flow data, water flow data, electricity flow data, inter-device electricity flow data, spatial location data, energy usage behavior data, questionnaire data, and historical operation strategy data;

[0013] preprocessing the collected multi-source data, and constructing a dissipative structure model, an ecological network model, a group behavior model, a spatiotemporal trajectory model, a thermodynamic constraint model, and a cultural evolution model based on the preprocessed multi-source data;

[0014] Based on the constructed model, quantitative indicators of each dimension are generated to form a six-dimensional evaluation system of self-organization characteristics, device collaboration, user behavior impact, spatiotemporal coupling characteristics, theoretical performance limits and operation strategy adaptability;

[0015] Based on the generated quantitative indicators of each dimension, a comprehensive analysis of the indicator data is performed;

[0016] The indicator data and model output data are processed to generate visual data.

[0017] The present invention has the following beneficial effects:

[0018] 1. This invention addresses the problem that traditional evaluation methods have a single dimension and cannot fully reflect the characteristics of the system. By constructing a six-dimensional evaluation system (self-organizing characteristics, equipment synergy, user group behavior impact, spatiotemporal coupling characteristics, theoretical performance limits, and operational strategy adaptability), based on advanced theories such as dissipative structure theory and ecological network analysis, it systematically evaluates the operating status of the integrated energy system. The solution generates 13 quantitative indicators (such as entropy production rate, reciprocity index, and comprehensive evaluation score), covering multiple aspects such as energy evolution, equipment interaction, and user behavior, solving the limitations of single-dimensional analysis. In order to address the problem of low data collection frequency and difficulty in capturing dynamic changes, this technical solution deploys multi-source sensors to collect data such as heat flow, gas flow, electricity flow, and user behavior in real time at a frequency of 1Hz, and transmits it to the central server via the AES-256 encrypted MQTT protocol. The model calculation module uses sliding window denoising and seasonal trend decomposition to ensure that the data accurately reflects seasonal fluctuations and emergencies (such as peak electricity consumption).

[0019] 2. This invention integrates dissipative structure theory, ecological network analysis, social physics, time geography, thermodynamic constraints, and cultural evolution theory to construct six evaluation models (dissipative structure model, ecological network model, etc.), deeply analyzing the system's energy evolution, device interaction, user behavior, and other characteristics. It also uses ARIMA, PageRank, TOPSIS, and SIR algorithms, combined with entropy and expert scoring to dynamically adjust indicator weights and generate an accurate comprehensive evaluation score (0-100). The verification module ensures the consistency of multi-source data through Pearson correlation analysis (high correlation r>0.7), and the anomaly detection mechanism (eliminating entropy production rates >2000 J / K·s) reduces the false positive rate.

[0020] 3. The present invention generates D3.js line charts, bar charts and spatial heat maps (100m×100m grids) through a visualization module to intuitively display the seasonal trends of indicators (such as the winter peak of entropy production) and the spatial distribution of resource allocation efficiency. The web interface supports interactive functions (time slider, area selection, indicator screening), and the report supports PNG, PDF, and HTML multi-format output, and provides an API interface to connect to third-party platforms. For example, in the case of the Southern Commercial Complex, the heat map identified inefficient areas (the extreme efficiency ratio increased from 0.4 to 0.55), and the interactive report shortened the analysis time from 10 minutes to 7 minutes (efficiency increased by 30%), and guided the optimization of equipment layout to improve resource efficiency by 15%. In the case of the suburban mixed park, the multi-format report met the needs of managers and engineers, and the comprehensive score increased by 8%, which significantly improved the practical application value of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a specific flow chart of the present invention;

[0022] Figure 2 It is the overall framework diagram of the present invention;

[0023] Figure 3 This is a simulation diagram for evaluating the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Specific embodiment 1

[0026] like Figure 1-3 As shown in Figure 1, a multi-dimensional quantitative evaluation method for the operation of a multi-season park integrated energy system systematically assesses the operational characteristics of the integrated energy system in different seasons by constructing a six-dimensional evaluation system. This system, centered on the coupling of dissipative structure theory and ecological network analysis, constructs a dynamic collaborative evaluation framework. Combining social physics, time geography, thermodynamic constraints, and cultural evolution theory, it comprehensively analyzes the system's self-organizing characteristics, equipment collaboration, the impact of user group behavior, spatiotemporal coupling characteristics, theoretical performance limits, and operational strategy adaptability. The entire evaluation process runs on a central server, which includes a model calculation module, an analysis module, a visualization module, and a verification module. These modules are responsible for model construction and indicator generation, comprehensive analysis, visualization report generation, and data cross-validation, respectively, ensuring that each module works together to produce reliable evaluation results.

[0027] The evaluation process starts with multi-source data collection, covering energy flow, spatial location, user behavior and historical policy data. In a typical park (such as an industrial or residential park with 1,000 households), about 50 temperature sensors (accuracy 0.1°C) are deployed at key nodes of heat flow such as heating pipelines and heat exchange stations to collect heat flow data and record temperature and time series information; 30 flow sensors (accuracy 1 liter / minute) are installed at key nodes of gas pipelines and water pipelines, such as the main pipeline entrance, to collect gas flow and water flow data; 40 voltage sensors (accuracy 0.01 kilowatt) are deployed at power transmission nodes such as substations and distribution boxes to collect power flow data; 100 smart meters record power generation data. The system collects data on power flow between energy devices such as generators, energy storage devices, and photovoltaic panels. Twenty GPS positioning devices (with an accuracy of 1 meter) are installed at energy devices (such as energy storage stations) and user activity areas to collect spatial location data. Approximately 1,000 user terminal devices (such as smart home systems and mobile applications) collect energy usage behavior data (such as daily electricity usage time and amount) and questionnaire data (such as energy usage preferences, collected quarterly via mobile applications). A database stores historical operational strategy data, covering scheduling parameters and strategy records for each season over the past three years, such as peak and valley load scheduling plans. These sensors and devices are connected to the central server via pre-set communication modules, covering approximately 90% of the park's energy nodes, ensuring comprehensive data collection.

[0028] Data is transmitted to the model calculation module of the central server via the MQTT protocol, encrypted with AES-256. Data on heat flow, gas flow, water flow, electricity flow, inter-device electricity flow, spatial location, and energy usage behavior are transmitted in real time at a 1Hz frequency to capture dynamic changes. Survey data is transmitted weekly, and historical operational strategy data is transmitted daily to accommodate its low-frequency update frequency. The MQTT protocol operates over a dedicated industrial-grade network with 100Mbps bandwidth. The central server is configured as a cluster capable of supporting 1TB of daily data throughput. A heartbeat mechanism (every 10 seconds) is used to detect transmission interruptions. If packet loss is detected (with a packet loss rate of less than 0.1%), data is automatically retransmitted. The central server uses the Hadoop Distributed File System (HDFS) to store data. Data is indexed by timestamp and device ID and sharded across at least three nodes, enabling fast retrieval and fault tolerance. If a single node fails, the system can fail over to a backup node within 1 minute.

[0029] After data is transferred to the model calculation module, it undergoes preprocessing and is used to construct six evaluation models corresponding to the analytical dimensions of the six-dimensional evaluation system. For heat flow, gas flow, and electricity flow data, the model calculation module uses sliding window denoising (5-second window size) and seasonal trend decomposition (24-hour period) to fill in no more than 5% of missing values. This generates standardized time series data for constructing a dissipative structure model, describing system evolution using nonlinear dynamic equations and outputting the system's entropy production rate and self-organizing state. For inter-device electricity flow, heat flow, and gas flow data, the module maps device IDs to network nodes and energy flows to directed edges (weighted by energy intensity), performing normalization to construct an ecological network model and analyze the energy interaction efficiency between devices. For user energy usage behavior data and questionnaire data, the module uses K-means clustering (setting five clusters based on Euclidean distance) to group user behaviors and conduct time series correlation analysis to construct a group behavior model reflecting user behavior patterns. Spatial location data is coordinate-normalized and, combined with the temporal distribution data of energy flow, is segmented into time windows to construct a spatiotemporal trajectory model for analyzing resource transfer efficiency. Energy balance calibration and entropy generation distribution calculations are performed on heat, gas, and electricity flow data to construct thermodynamic constraint models and assess performance limits. Historical operational strategy data is used to calculate strategy parameter entropy values ​​and conduct time series analysis, constructing a cultural evolution model and analyzing strategy adaptability. The construction of these models is performed by the model calculation module.

[0030] Based on the above model, the model calculation module generates quantitative indicators for each dimension. Through time series analysis based on the ARIMA model and seasonal differencing, the entropy production rate and self-organization efficiency are calculated to reflect the proportion of orderly energy utilization in the system. Furthermore, the dissipation gradient index is calculated based on the slope of the entropy production rate change in different seasons to characterize the system's responsiveness to seasonal changes. A PageRank-based graph theory algorithm is used to calculate the reciprocity index, and a closed-loop path analysis algorithm is used to calculate the circulation rate, reflecting the intensity of energy interaction between devices and the proportion of energy reuse, respectively. Behavioral consistency and system responsiveness indicators are calculated using a statistical method based on principal component analysis to analyze the impact of user behavior on the system. Path analysis based on the Dijkstra algorithm, combined with weighted distance calculation, calculates spatiotemporal reachability and time compression rate, reflecting resource transfer efficiency and operational flexibility. Energy balance analysis based on the Carnot cycle calculates the limiting efficiency ratio and loss concentration to identify efficiency bottlenecks. Through dynamic propagation analysis based on the SIR model and dynamic adjustment of policy adoption rates, strategy diversity and propagation efficiency are calculated to assess strategy adaptability.

[0031] After each dimension's indicators are generated, they are transmitted hourly from the model calculation module to the analysis module via the AES-256 encrypted MQTT protocol. The analysis module uses a TOPSIS-based multi-objective decision analysis algorithm, combined with entropy and expert scoring, to determine the weights of each indicator. This module then performs weighted normalization on the indicator data to generate a comprehensive evaluation score reflecting the system's dynamic performance and multi-seasonal adaptability. Simultaneously, the indicator data and model output data are transmitted at the same frequency to the visualization module, which performs time series interpolation and spatial gridding on the data to generate visualizations. D3.js-based charts display seasonal trends in the indicators, and spatially interpolated heat maps demonstrate the spatial distribution of resource allocation efficiency. Finally, a visual report containing each evaluation indicator is generated, showcasing the system's operational characteristics and optimization potential.

[0032] To ensure the reliability of the evaluation results, evaluation data is transmitted hourly from the model calculation module to the verification module via the MQTT protocol, which uses AES-256 encryption. The verification module compares data generated from different evaluation directions (such as the correlation between spatiotemporal accessibility and the reciprocity index) using a Pearson correlation coefficient analysis. It then calculates the correlation matrix and performs significance tests to generate a consistency analysis report reflecting the robustness of the system's operational characteristics. Collaborative analysis and data cross-validation across modules ensure that the evaluation system comprehensively reflects the system's dynamic performance and multi-season adaptability, generating the final multi-season operational characteristics analysis results. Specific embodiment 2

[0034] like Figure 1-3 As shown, the key algorithms mentioned in Example 1 are analyzed in detail below, including their core mathematical formulas and explanations:

[0035] ARIMA model (time series analysis):

[0036] The core formula is as follows:

[0037] y t =c+φ1y t-1 +φ2y t-2 +…+φ p y t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t

[0038] in:

[0039] y t : Observed value at time t (such as energy flow data); c: constant term; φ i: Autoregressive (AR) coefficient, p is the autoregressive order; θ j : Moving average (MA) coefficient, q is the moving average order; ∈ t : Random error term; Seasonal difference: Δ s y t =y t -y t-s , s is the seasonal period (e.g. 24 hours).

[0040] The ARIMA model combines autoregression, differencing, and moving averages to address trends and seasonal fluctuations in time series data. It uses an ARIMA (1,1,1) model (p=1, d=1, q=1) combined with seasonal differencing (24-hour period) to analyze time series data on heat, gas, and electrical energy flows, calculating entropy production (the rate at which the system dissipates energy), self-organization efficiency (the proportion of orderly energy utilization), and dissipation gradient (the ability to respond to seasonal changes).

[0041] During energy system operation, energy flow data (such as heat flow) is affected by seasonal variations (e.g., high heating demand in winter). Traditional static analysis cannot capture dynamic dissipation characteristics, resulting in distorted evaluation results. For example, fluctuations in heat flow in winter can mask system efficiency issues.

[0042] The core formula of the PageRank algorithm (graph theory algorithm) is as follows:

[0043]

[0044] Where: PR(o): PageRank value (reciprocity index) of node i (energy device); d: damping factor (set to 0.85 in the working mode); N: total number of nodes (number of devices); M(i): set of nodes pointing to node i; L(j): number of outgoing edges of node j (number of energy flow outputs).

[0045] The PageRank algorithm iteratively calculates the importance of nodes in the network to assess the strength of energy interactions between devices. In its working mode, the ecological network model uses devices as nodes and energy flows as edges. PageRank calculates the reciprocity index, which reflects the collaborative importance of devices in the energy interaction network.

[0046] Energy interactions between devices (such as energy storage devices transmitting power to the grid) vary in intensity, making it difficult to quantify which devices contribute most to the collaborative effort. This leads to inefficient resource allocation. For example, a particular energy storage device may interact frequently but contribute little, impacting the overall collaborative nature of the system.

[0047] Solution:

[0048] Formula Application: The model calculation module constructs a directed graph based on data on electrical energy flow, thermal energy flow, and gas flow between devices, with edge weights representing energy intensity. PageRank iteratively calculates the reciprocity index for each device, with higher values ​​indicating a greater importance of the device in the interaction.

[0049] Implementation: Taking a network of 100 devices as an example, the module processes hourly energy flow data, constructs an adjacency matrix (100×100), initializes the PageRank value of each node to 1 / 100, and iterates 10 times (convergence error <0.01). The reciprocity index is the normalized PageRank value, ranging from 0 to 1.

[0050] Solution effect: It solved the problem of difficult to quantify the importance of equipment collaboration, identified key equipment (such as energy storage stations with a mutual benefit index > 0.8), and guided resource allocation optimization (such as prioritizing high-mutual benefit equipment to reduce energy loss by 10%).

[0051] The core formula of the closed-loop path analysis algorithm is:

[0052]

[0053] Where: CR: circulation rate (proportion of energy reuse paths); C: the set of closed-loop paths in the network (at least three nodes); w(c): the total weight of closed-loop path c (energy flow intensity); E: the set of all edges in the network; w(e): the weight of edge e (single energy flow intensity).

[0054] The closed-loop path analysis algorithm detects loop paths in the network and calculates the energy reuse ratio. In operation, the ecological network model uses this algorithm to calculate the recycling rate, which reflects the efficiency of energy recycling between devices.

[0055] Actual problem:

[0056] In energy systems, some energy (such as waste heat) can be reused through recycling paths. However, traditional analysis makes it difficult to quantify recycling efficiency, resulting in insufficient waste heat recovery. For example, a heating system may waste 20% of thermal energy due to a lack of recycling paths.

[0057] Solution:

[0058] Formula application: The model calculation module detects closed-loop paths with more than three nodes (such as generator → energy storage → load → generator) based on energy flow data between devices, and calculates the proportion of the total energy flow in the closed-loop path to the total energy flow.

[0059] Implementation: Using hourly energy flow data as an example, the module constructs a directed graph and uses depth-first search (DFS) to identify closed-loop paths. The energy intensity of these closed-loop paths (in MW) is accumulated and divided by the total edge energy intensity. The recycling rate ranges from 0 to 1, with higher values ​​indicating greater reuse efficiency.

[0060] Solution effect: It solves the problem of difficult to quantify recycling efficiency, identifies efficient recycling paths (such as waste heat recovery paths with a circulation rate > 0.5), and guides the deployment of waste heat recovery equipment (such as adding heat exchangers to increase thermal energy utilization by 15%).

[0061] The core formula of dynamic regression algorithm:

[0062] SR=β1x1+β2x2+…+β n x n

[0063] Where: SR: system responsiveness index (the degree of system response to behavioral fluctuations); β i : regression coefficient (estimated by least squares method); x i : User behavior fluctuation characteristics (such as standard deviation of electricity consumption).

[0064] Dynamic regression uses a linear model to quantify the impact of an independent variable (user behavior) on a dependent variable (system operating parameters). In operation, the algorithm calculates a system responsiveness index, which reflects the system's sensitivity to fluctuations in user behavior.

[0065] Practical Problem: Fluctuations in user behavior (such as sudden increases in power usage) can cause system instability, making it difficult to quantify system responsiveness through traditional analysis. For example, a sudden peak in power usage can cause a 10% efficiency loss.

[0066] Solution:

[0067] Formula application: The model calculation module uses user behavior fluctuations (such as the daily standard deviation of electricity consumption) as independent variables and system operation parameters (such as load rate) as dependent variables to fit the regression model. The system responsiveness is the weighted average of the regression coefficients.

[0068] Implementation: Using daily electricity consumption data from 1,000 users and the system load factor as an example, the module constructs a regression model with the 24-hour standard deviation of electricity consumption as input and the change in load factor as output. The regression coefficient is estimated (with a significance level of p < 0.05). The system response is the mean of the coefficient, with dimensionless units.

[0069] Solution: This solves the problem of difficult-to-quantify system response capabilities, identifies high-response scenarios (such as nighttime peaks with a response capability > 1), and guides dynamic scheduling (such as increasing energy storage discharge to reduce the impact of 20% fluctuations). The core formula of the SIR model (propagation dynamic analysis):

[0070]

[0071] SD=H(p),

[0072] Where: S, I, R: number of strategies in susceptible, infected, and recovered states; β: infection rate (set to 0.1 in the working mode); γ: recovery rate (set to 0.05); SD: strategy diversity (entropy value of strategy parameters, H(p) = -∑p i logp i ); SE: dissemination efficiency (final adoption rate, R ∞ is the final number of adopted strategies, and N is the total number of strategies).

[0073] The SIR model simulates the dynamics of strategy propagation in a system. In its working mode, the algorithm analyzes historical strategy data and calculates strategy diversity (the diversity of strategy parameters) and propagation efficiency (the proportion of strategy adoption).

[0074] Actual problem:

[0075] The diversity and adoption efficiency of operational strategies (such as peak-valley scheduling) are difficult to quantify, and traditional analysis cannot evaluate strategy adaptability. For example, a low-diversity strategy may result in a 10% efficiency loss.

[0076] Solution:

[0077] Formula application: The model calculation module regards strategies as "individuals", simulates their adoption process, and calculates diversity (entropy value) and dissemination efficiency (adoption rate).

[0078] Implementation: Using 100 historical strategies as an example, the module constructs a strategy parameter vector (dispatching time, load distribution) and calculates entropy (strategy diversity). The SIR model uses daily adoption rate data as input and iterates 100 steps (one-day time step). The propagation efficiency is the proportion of strategies ultimately adopted, ranging from 0 to 1.

[0079] Solution effect: It solved the problem of difficult to quantify strategy adaptability, identified high-efficiency strategies (such as dynamic scheduling with a propagation efficiency > 0.8), and guided strategy promotion (such as promoting high-adoption rate strategies to improve operational efficiency by 20%). Specific embodiment 3

[0081] like Figure 1-3 As shown, the following is a description of the specific application logic steps of each module and algorithm in the multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system:

[0082] The model calculation module is the core of the evaluation system, responsible for processing multi-source data, constructing six evaluation models, and generating quantitative indicators for the six-dimensional evaluation system. It receives data transmitted via the AES-256 encrypted MQTT protocol, including data collected at a 1Hz frequency on heat flow, gas flow, water flow, electricity flow, inter-device electricity flow, spatial location and energy usage behavior, weekly questionnaire data, and daily historical operation strategy data. Data is stored in the Hadoop Distributed File System (HDFS), indexed by timestamp and device ID, and sharded across three nodes to ensure a packet loss rate of less than 0.1%. Any missing packets trigger retransmission. The module first preprocesses the data: heat, gas, and electricity flow data are denoised using a sliding window (5-second window size) and seasonal trend decomposition (24-hour period), with no more than 5% missing values ​​filled in to generate normalized time series. Inter-device energy flow data are mapped into a node-edge structure (device ID as node, energy intensity as edge weight) and normalized to the range 0-1. User energy usage behavior and questionnaire data are grouped using K-means clustering (5 clusters, based on Euclidean distance), and time series correlation is calculated. Spatial location data are normalized to the WGS84 coordinate system and segmented into 1-hour windows based on the temporal distribution of energy flow data. Scheduling parameters are extracted from historical operation strategy data, and Shannon entropy is calculated. This preprocessed data is used to construct dissipative structure models, ecological network models, group behavior models, spatiotemporal trajectory models, thermodynamic constraint models, and cultural evolution models. Model construction runs on a GPU cluster, taking approximately 10 minutes per run. Based on these models, the module executes the following algorithm to generate 13 metrics: entropy production rate, self-organization efficiency, dissipation gradient, reciprocity index, cycle rate, behavioral consistency, system responsiveness, spatiotemporal reachability, time compression rate, limit efficiency ratio, loss concentration, strategy diversity, and propagation efficiency. Once generated, these metrics are transmitted hourly via the MQTT protocol to the analysis, visualization, and verification modules, along with a timestamp and metric ID.

[0083] The ARIMA model is used to quantify self-organizing properties, addressing the difficulty in capturing dynamic dissipation characteristics due to seasonal fluctuations in energy flows. Using standardized time series of heat, gas, and electricity flows (24-hour window, approximately 86,400 points) as input, the module applies seasonal differencing (24-hour period) to remove diurnal fluctuations and generate stationary series. An ARIMA (1,1,1) model (autoregressive order 1, differencing order 1, and moving average order 1) is configured to fit the series using maximum likelihood estimation to predict energy flow trends. The entropy production rate is calculated based on the statistical mean of the fitted series, reflecting the energy dissipation rate (units: J / K·s). The self-organizing efficiency is the proportion of the stationary (low-variance) portion of the fitted series, ranging from 0 to 1. The dissipation gradient is calculated as the difference between the slopes of the spring and winter fitted series, indicating seasonal responsiveness. Output indicators are transmitted to the analysis module to address low winter heating efficiency (e.g., less than 10% of summer efficiency) and guide scheduling optimization (e.g., increasing heating power).

[0084] The PageRank algorithm quantifies device synergy, addressing the uneven intensity of energy interactions between devices and the difficulty in identifying key devices. Using hourly data on the flow of electricity, heat, and gas between devices as input, the module constructs a directed graph (100 devices, 100×100 adjacency matrix), with edge weights representing energy intensity. The PageRank value is initialized to 1 / 100, the damping factor is set to 0.85, and 10 iterations are performed (convergence error <0.01). The reciprocity index is calculated as a normalized PageRank value (range 0-1), reflecting the importance of device interactions. The output reciprocity index is transmitted to the analysis module to identify highly reciprocal devices (such as energy storage stations with an index >0.8) and optimize resource allocation (such as priority scheduling to reduce losses by 10%).

[0085] The closed-loop path analysis algorithm further quantifies the synergy of equipment and solves the problem of low energy recycling efficiency. Taking the hourly energy flow data between devices as input, the module constructs a directed graph with the edge weight as energy intensity (unit MW). Through depth-first search (DFS), closed-loop paths with more than three nodes (such as generator → energy storage → load → generator) are detected, and the circulation rate is calculated as the sum of the energy intensities of the closed-loop paths divided by the total edge energy intensity (range 0-1). The output circulation rate is transmitted to the analysis module to identify efficient circulation paths (such as waste heat recovery paths with a circulation rate > 0.5) and guide equipment deployment (such as adding heat exchangers to increase thermal energy utilization by 15%).

[0086] Principal component analysis (PCA) quantifies the impact of user group behavior, addressing the challenge of large differences in user behavior and the difficulty in assessing consistency. Taking the daily electricity usage data of 1,000 users (24-hour series) as input, the module constructs a 1,000-by-24 matrix, calculates the covariance matrix after standardization, and extracts the first three principal components (explaining 80% of the variance). Behavioral consistency is calculated as the sum of the first three eigenvalues ​​divided by the sum of the total eigenvalues ​​(range 0-1). This is output to the analysis module, which identifies user groups with high consistency (e.g., residential areas with consistency > 0.7) and optimizes load scheduling (e.g., centralized scheduling to reduce peak load by 15%).

[0087] The dynamic regression algorithm further quantifies the impact of user group behavior, addressing the issue of system instability caused by user behavior fluctuations. Using user behavior fluctuations (24-hour standard deviation of electricity consumption) and system operating parameters (such as load factor) as input, the module constructs a regression dataset (1000 users x 24 hours), fits a linear regression model, and estimates the regression coefficient (significance p < 0.05). System responsiveness is the dimensionless mean of the regression coefficient, which is output to the analysis module. This module identifies high-responsiveness scenarios (such as nighttime peaks with responsiveness > 1) and optimizes dynamic scheduling (such as increasing energy storage discharge to reduce the impact of 20% fluctuations).

[0088] The Dijkstra algorithm quantifies spatiotemporal coupling, addressing the difficulty of assessing resource transfer efficiency due to spatial and temporal constraints. Using spatial location and hourly energy flow temporal distribution data as input, the module constructs a weighted graph (100 devices, with edge weights of 0.5 × distance + 0.5 × delay, measured in minutes) and calculates the shortest path from a starting point (e.g., a power station) to each node. Spatiotemporal accessibility, expressed as the average path length (in minutes), is output to an analysis module, which identifies areas of low accessibility (e.g., edge devices with accessibility > 30 minutes) and optimizes the layout (e.g., adding relay stations to reduce transfer time by 15%).

[0089] The time series compression algorithm further quantifies the spatiotemporal coupling characteristics to address the issues of excessively long scheduling windows and low flexibility. Using energy flow temporal distribution data (24-hour windows) as input, the module uses cluster analysis (with a 1-hour window size) to identify low-load periods (e.g., 2:00 AM to 4:00 AM) and compress redundant time periods (e.g., from 24 hours to 20 hours). The time compression ratio, a ratio of compressed duration (ranging from 0 to 1), is output to the analysis module, which identifies compressible periods (e.g., nighttime windows with a compression ratio > 0.2) and optimizes scheduling (e.g., shortening the cycle time to improve resource utilization by 10%).

[0090] Carnot cycle analysis quantifies theoretical performance limits, solving the problem of efficiency being constrained by thermodynamics and bottlenecks being difficult to locate. Using hourly heat flow, gas flow, and electrical energy flow data as input, the module calibrates the energy balance (error < 1%) and calculates the actual efficiency (output heat / input fuel) and the Carnot efficiency (based on pipe temperatures, such as 373K and 293K). The limiting efficiency ratio is the ratio of the actual efficiency to the Carnot efficiency (range 0-1), and the loss concentration is the spatial standard deviation of entropy production (unit J). 2 / K 2 The output indicators are transmitted to the analysis module to identify low-efficiency areas (such as pipeline sections with efficiency ratios < 0.5) and optimize equipment (such as replacing a heat exchanger to increase efficiency by 10%).

[0091] The SIR model quantifies the adaptability of operational strategies, addressing the difficulty in assessing strategy diversity and adoption efficiency. Taking 100 historical operational strategy data points (scheduling parameters) as input, the module calculates the Shannon entropy of the parameters and constructs a time series of adoption rates. The SIR model is configured (infection rate 0.1, recovery rate 0.05), iterates 100 times (1 day), and calculates strategy diversity as parameter entropy (in bits) and propagation efficiency as the final proportion of adopted strategies (ranging from 0 to 1). The output metrics are transmitted to the analysis module, which identifies high-efficiency strategies (e.g., dynamic scheduling with a propagation efficiency > 0.8) and promotes these strategies (e.g., improving operational efficiency by 20%).

[0092] The analysis module integrates multi-dimensional indicators to generate a comprehensive evaluation score, addressing the difficult balance between indicator weights and comprehensiveness. It receives indicator data transmitted hourly by the model calculation module, stores it in a temporary cache, verifies its integrity, and then linearly normalizes it (mapping it to a range of 0-1). The module uses the entropy method to calculate the information entropy of each indicator and determines its weight (ranging from 0-1, summing to 1). The weights are adjusted based on the Delphi scores of five experts (e.g., a weight of 0.2 for entropy production). Using the TOPSIS algorithm, the module constructs a normalized matrix (1000 time points × 12 indicators), identifies positive ideal solutions (maximum values ​​for each indicator) and negative ideal solutions (minimum values), calculates the Euclidean distance from each time point to the positive and negative ideal solutions, and generates a comprehensive evaluation score (proximity, 0-100). The module analyzes score trends and generates operational scheduling recommendations (e.g., increasing load during high-scoring periods), spatial layout recommendations (e.g., adding equipment in low-scoring areas), and long-term strategic recommendations (e.g., promoting scheduling patterns during high-scoring seasons). The comprehensive evaluation scores and analysis results are transmitted to the visualization module to solve the multi-dimensional comprehensive evaluation problem and guide optimization (such as adjusting the scheduling in low-scoring seasons to improve the overall performance by 15%).

[0093] The visualization module generates intuitive evaluation reports to solve the problem of difficult presentation of operating characteristics. It receives the indicator data transmitted every hour by the model calculation module, stores it in HDFS, and then performs time series interpolation (linear interpolation, step size 1 hour) and spatial gridding (100m×100m grid) to generate visual data. The module uses D3.js to generate line charts and bar charts to show the seasonal trend of indicators (such as the winter peak of entropy production rate), and generates heat maps through Kriging interpolation method to show the spatial distribution of resource allocation efficiency (such as high-efficiency areas are concentrated in the center of the park). The report integrates charts and heat maps, outputs them through the web interface, highlights optimization potential (such as areas with a 10% reduction in peak load), and stores them in HDFS for managers to access and guide optimization (such as adjusting equipment layout to improve resource efficiency by 15%).

[0094] The validation module improves the reliability of results through data cross-validation, addressing the issue of inconsistent results across different evaluation dimensions. It receives indicator data transmitted hourly by the model calculation module (priority higher than visualization data, QoS1 level), constructs an indicator matrix (1000 time points × 12 indicators), calculates the Pearson correlation coefficient for each pair of indicators (such as spatiotemporal accessibility and reciprocity index, significance level p < 0.05), and generates a correlation matrix. Highly correlated pairs (r > 0.7) are marked as highly consistent, while low-correlation pairs are recommended for further analysis. The module aggregates correlation coefficients and p-values ​​to generate a consistency report, highlighting robust features (such as accessibility-synergy correlation > 0.7). This report is stored in HDFS and transmitted to the visualization module for inclusion in the comprehensive report, addressing evaluation accuracy issues (e.g., reducing false positives by 10%). Specific embodiment 4

[0096] like Figure 1-3 As shown, the following are specific use cases of this technical solution:

[0097] Case 1: Winter operation optimization of a large industrial park in northern China:

[0098] A large industrial park in northern China, housing numerous energy-intensive enterprises such as steel and chemical companies, has large and complex energy demands. The park deployed 50 temperature sensors at key nodes in heating pipelines, 30 flow sensors in gas and water pipelines, and 40 voltage sensors at power transmission nodes. 100 smart meters record the flow of energy between energy devices, 20 GPS tracking devices collect the spatial locations of energy devices and users, and 1,000 user terminals collect data on energy usage.

[0099] After data collection, it is transmitted to a central server at a specified frequency. The model calculation module performs time series preprocessing on heat, gas, and electricity flow data to construct a dissipative structure model. This model found that increased production demand in winter increases the system's entropy production rate, decreases its self-organizing efficiency, and demonstrates that the system's response to seasonal changes is insufficient. The reciprocity index and circulation rate calculated using the ecological network model reveal low energy interaction efficiency between some devices and insufficient energy reuse. The analysis module combines the TOPSIS algorithm and entropy method to determine the weights for each indicator, resulting in a low overall evaluation score.

[0100] Based on this, the park adjusted its energy equipment operation strategy, optimized the energy exchange paths between devices, and improved energy reuse efficiency. During the re-evaluation, the overall score improved significantly, and the system operated more efficiently and stably.

[0101] Case 2: Summer energy-saving renovation of a commercial complex in the south:

[0102] In a commercial complex in southern China, air conditioning and other cooling equipment account for a significant portion of energy consumption in the summer. Sensors and equipment were deployed at energy flow nodes to collect data. The model calculation module constructed a group behavior model. By analyzing user energy usage data, it was found that most users frequently used air conditioning during daytime office hours and set the temperature to a lower setting. A spatiotemporal trajectory model revealed that energy transfer in some areas was subject to time delays and long paths. The thermodynamic constraint model calculated a low limit efficiency ratio, high loss concentration, and an efficiency bottleneck in the cooling system.

[0103] After the analysis module generated a comprehensive evaluation score, the commercial complex conducted energy-saving publicity campaigns targeting user behavior, encouraging rational use of air conditioners. Simultaneously, the spatial layout of energy equipment was adjusted, energy transmission paths were optimized, and the refrigeration system was upgraded. After a period of operation, further evaluation revealed improved system consistency, enhanced spatial and temporal accessibility, and a higher maximum efficiency ratio, achieving the energy conservation and consumption reduction goals.

[0104] Case 3: Optimization of seasonal energy scheduling strategies for a suburban mixed park:

[0105] Suburban mixed-use parks contain industrial enterprises, residential buildings, and commercial facilities, and their energy demand varies significantly with the seasons. Through long-term data collection and model calculations, the cultural evolution model analyzed historical operational strategy data and found that the existing energy scheduling strategy lacked seasonal adaptability, had low strategy diversity, and exhibited poor dissemination efficiency. Combined with other dimensional indicators, the analysis module generated a comprehensive evaluation score, clarifying the system's strengths and weaknesses across different seasons.

[0106] Based on the evaluation results, the park developed differentiated energy scheduling strategies for each season. In spring and autumn, the operating power of some equipment was reduced to optimize energy distribution. In summer and winter, the coordinated operation of energy equipment was adjusted to address peak loads, introducing new energy-saving strategies. After implementing the new strategies, further evaluation revealed improved strategy diversity and dissemination efficiency, significantly enhancing the system's multi-season adaptability and effectively improving energy efficiency.

[0107] The following is a supplementary explanation of the "Quantitative Evaluation Method for Multi-Dimensional Operation of Multi-Seasonal Campus Integrated Energy Systems," covering data anomaly handling, computing resource requirements, algorithm parameter optimization, dynamic weight adjustment, visual interactivity, verification of anomaly detection, case quantitative results, and solution scalability and standardization, aiming to improve the practicality and universality of the technical solution. Specific embodiment 5

[0109] like Figure 1-3 As shown, the following is a supplementary explanation of the "Multi-dimensional Operation Quantitative Evaluation Method for Multi-season Campus Integrated Energy System", covering data anomaly processing, computing resource requirements, algorithm parameter optimization, dynamic weight adjustment, visual interactivity, verification of anomaly detection, case quantitative results, and solution scalability and standardization, aiming to improve the practicality and universality of the technical solution.

[0110] Exception handling of data collection:

[0111] To ensure the reliability of data collection, the system has designed a multi-level exception handling mechanism:

[0112] Sensor fault detection: If a temperature sensor, flow sensor, or voltage sensor does not return data for five consecutive seconds, the central server triggers an alarm through the MQTT protocol, notifying maintenance personnel to inspect the equipment (such as replacing the faulty sensor) and switch to the backup sensor (one backup sensor is deployed at each key node, with a coverage rate of 10%).

[0113] Data cleaning rules: The model calculation module detects anomalies in the collected data, eliminating values ​​exceeding three standard deviations (e.g., thermal energy flow temperatures >500K or <200K are considered invalid) and using linear interpolation to fill in outliers (no more than 5% of the data volume). Negative data (e.g., power flow <0) is automatically marked as 0, and an anomaly log is recorded.

[0114] Network interruption handling: If MQTT transmission is interrupted (heartbeat detection does not receive a response every 10 seconds), the sensor device caches the data in local storage (1GB capacity, supporting approximately one hour of data) and retransmits it in batches after the network is restored. The central server monitors the packet loss rate and automatically reduces the transmission frequency (for example, from 1Hz to 0.5Hz) if it exceeds 0.1%, prioritizing the transmission of critical data (such as heat and power flow).

[0115] Effect: The exception handling mechanism improves data integrity from 95% to 99%, reducing evaluation bias caused by sensor failure or network interruption (such as a 5% miscalculation of entropy production).

[0116] Computational resource requirements for model building:

[0117] The model calculation module runs in a high-performance computing environment, ensuring the rapid construction of six evaluation models (dissipative structure, ecological network, etc.):

[0118] Hardware Configuration: The module is deployed in a GPU cluster equipped with four NVIDIA A100 GPUs (40GB of video memory), 128GB of DDR4 memory, and a 2TB NVMe SSD. It supports processing one hour of data (approximately 1GB) from 1,000 campuses at a time. The CPU is an AMD EPYC 32-core processor to accelerate data preprocessing.

[0119] Parallel computing framework: Using the Apache Spark framework, data preprocessing (denoising, clustering) and model building (ARIMA, PageRank) are distributed to multiple nodes for parallel execution. Each model is allocated two GPU cores, and a single build takes approximately 10 minutes, with a peak throughput of 1TB / hour.

[0120] Load balancing: Through Spark's dynamic resource allocation, the module adjusts the number of computing nodes (3-5 nodes) based on the data volume to avoid overloading a single node. If GPU utilization exceeds 90%, low-priority tasks (such as questionnaire data processing) are automatically suspended, and 1Hz energy flow data is prioritized.

[0121] Results: The high-performance computing environment shortens model building time from 15 minutes to 10 minutes, supports data processing for large-scale campuses (e.g., 5,000 households), and improves computing efficiency by 30%.

[0122] Algorithm parameter optimization:

[0123] To ensure algorithm performance, the parameters of each key algorithm are determined through systematic tuning:

[0124] ARIMA parameter selection: The model calculation module uses the Akaike Information Criterion (AIC) to evaluate the ARIMA model. We test order combinations (p, d, q) from (0, 0, 0) to (2, 2, 2), selecting the ARIMA configuration (1, 1, 1) with the lowest AIC as the default. For seasonal data (such as high volatility in winter), we dynamically adjust the differencing order d (1 or 2) to ensure series stationarity.

[0125] PageRank Iteration Optimization: The PageRank algorithm was set to a damping factor of 0.85, and based on convergence analysis (error < 0.01), the number of iterations was determined to be 10. For small campuses (<50 devices), this number was reduced to 5 iterations, reducing the computational load by 20%. For large campuses (>200 devices), this number was increased to 15 iterations to ensure accuracy.

[0126] SIR model parameter adjustment: The infection rate β (0.1) and recovery rate γ (0.05) of the SIR model were optimized through grid search, testing β∈[0.05,0.2] and γ∈[0.02,0.1], selecting the parameter combination that maximizes transmission efficiency. For scenarios with a large number of strategies (>200), the number of iterations was increased to 150.

[0127] Effect: Parameter optimization improves the algorithm accuracy by 5% (e.g., the entropy production error is reduced from ±50 J / K·s to ±30 J / K·s), adapting to different park sizes and seasonal characteristics.

[0128] Dynamic adjustment of indicator weights:

[0129] The analysis module supports dynamic adjustment of indicator weights to adapt to different seasons and park types:

[0130] Seasonal Weight Adjustment: The module adjusts weights based on seasonal characteristics. For example, in winter, when thermal energy flows dominate, the weights of entropy production rate and limit efficiency ratio are increased by 10% (e.g., from 0.2 to 0.22); in summer, when electrical energy flows dominate, the weights of spatiotemporal accessibility and behavioral consistency are increased by 10%. This adjustment is based on changes in information entropy based on historical data (triggered by an entropy difference > 0.1).

[0131] Park type adaptation: Industrial parks (such as in Case 1) prioritize equipment synergy (weights of the reciprocity index and recycling rate increased to 0.25), while commercial parks (such as in Case 2) prioritize user behavior (weights of behavioral consistency and system responsiveness increased to 0.25). Adjustments are made through a combination of expert scoring and entropy analysis, with expert scoring updated quarterly.

[0132] Dynamic update process: The analysis module analyzes the distribution of indicator data every month. If the volatility of a certain indicator increases (such as standard deviation > 20%), the weight is recalculated using the entropy method and combined with the quick scoring of three experts (Delphi method, completed in 1 hour) to update the weight table.

[0133] Effect: Dynamic weight adjustment increases the seasonal adaptability of the comprehensive evaluation score by 10% (for example, the winter score increases from 80 to 88), enhancing its pertinence in different scenarios.

[0134] Interactivity and output formats of visual reports:

[0135] The visualization module provides interactive reports and multi-format output to meet different user needs:

[0136] Interactive Features: The web interface supports a time slider (adjusting the time range, such as a single day or an entire season), region selection (zooming in on a 100m x 100m grid area), and metric filtering (displaying entropy production alone or a composite score). Users can click on a heat map node to view detailed data (such as the loss concentration in a specific area).

[0137] Multi-format output: Reports can be exported in PNG (high resolution, 1920×1080), PDF (A4 format, with embedded charts and annotations), and HTML (interactive web pages, including D3.js code). The module provides an API interface, allowing third-party systems (such as energy management platforms) to access visualization data.

[0138] Customization support: Users can set chart styles (such as color and font), heat map thresholds (such as efficiency > 0.8 is marked green) through the web interface, and save templates for subsequent use.

[0139] Results: Interactive reports improve user efficiency by 30% (e.g., analysis time is reduced from 10 minutes to 7 minutes), and multi-format output meets the needs of different roles such as managers and engineers.

[0140] Anomaly detection of verification module:

[0141] The verification module enhances consistency analysis and handles low correlation and abnormal indicators:

[0142] Low correlation handling: If the Pearson correlation coefficient r is less than 0.3 (e.g., spatiotemporal accessibility and reciprocity index), the module triggers a second validation: recollecting one hour of data and calibrating model parameters (e.g., Dijkstra weights). If the correlation remains low, the analysis module is advised to adjust the indicator weights or add additional data dimensions (e.g., including ambient temperature).

[0143] Abnormal indicator alarm: If an indicator shows a negative correlation (r<0) or an abnormal value (such as entropy production rate >2000 J / K·s, which is three times higher than the historical mean), the module marks it as abnormal, suspends transmission to the analysis module, and triggers an alarm to the administrator, suggesting that the data source (such as sensor failure) or model configuration be checked.

[0144] Verification report optimization: A visual table is added to the consistency report to highlight high correlation pairs (r>0.7, marked in green) and low correlation pairs (r<0.3, marked in red), and comes with suggestions (such as "Recommendation for calibrating Dijkstra algorithm parameters").

[0145] Effect: Anomaly detection reduces the misjudgment rate of evaluation results by 10% (for example, the misidentification rate of high-efficiency devices is reduced from 5% to 0.5%), improving the robustness of the system.

[0146] Quantitative results of the case:

[0147] To enhance the persuasiveness of the case, the following provides a quantitative comparison of key indicators and comprehensive scores in the three cases:

[0148] Case 1 (Northern Industrial Park, Winter):

[0149] Before optimization: entropy production rate 1200 J / K·s, self-organization efficiency 0.65, reciprocity index 0.6, circulation rate 0.4, and comprehensive score 75.

[0150] After optimization: entropy production rate 1000 J / K·s (decreased 17%), self-organization efficiency 0.78 (increased 20%), reciprocity index 0.75 (increased 25%), circulation rate 0.55 (increased 38%), and comprehensive score 85 (increased 13%).

[0151] Measures: Optimize energy interaction paths and add waste heat recovery equipment.

[0152] Case 2 (Southern Commercial Complex, Summer):

[0153] Before optimization: behavioral consistency 0.5, spatial and temporal reachability 40 minutes, extreme efficiency ratio 0.4, and comprehensive score 70.

[0154] After optimization: behavioral consistency 0.7 (increased 40%), time and space accessibility 30 minutes (decreased 25%), extreme efficiency ratio 0.55 (increased 38%), and comprehensive score 82 (increased 17%).

[0155] Measures: Energy-saving publicity, optimization of transmission paths, and upgrading of refrigeration systems.

[0156] Case 3 (urban-suburban mixed park, four seasons):

[0157] Before optimization: strategy diversity 2.0 bits, communication efficiency 0.6, comprehensive score (spring / summer / autumn / winter) 78 / 75 / 77 / 74.

[0158] After optimization: strategy diversity 2.8bit (increased by 40%), communication efficiency 0.8 (increased by 33%), comprehensive score (spring / summer / autumn / winter) 85 / 83 / 84 / 81 (average increase of 8%).

[0159] Measures: Differentiated scheduling and promotion of high-efficiency strategies.

[0160] Effect: Quantitative results intuitively demonstrate optimization effectiveness, enhance the credibility of technical solutions, and guide managers to make accurate decisions.

[0161] Scalability and standardization of solutions:

[0162] The technical solution is designed with scalability and standardization mechanisms to adapt to parks of different sizes and types:

[0163] Scale adaptation:

[0164] Small campus (500 households): Reduce the number of sensors (20 temperature sensors, 10 flow sensors), lower the model calculation complexity (such as 5 PageRank iterations), and run it on a single-node server (16GB of memory).

[0165] Large campuses (5,000 households): Increase the number of sensors (200 temperature sensors and 100 flow sensors), adopt a multi-node GPU cluster (8 A100 GPUs), and support 10TB / day data throughput.

[0166] Modular design: The model calculation module supports dynamic expansion. Adding new dimensions (such as environmental impact) only requires adding models and indicators, and deployment time is less than 1 week.

[0167] Type adaptation:

[0168] Hospital campus: Added medical equipment energy consumption data, adjusted weights (behavior consistency weight increased to 0.3), and optimized real-time response (such as accessibility <10 minutes).

[0169] Campuses: Incorporating student behavior data (such as dormitory electricity usage) optimizes nighttime scheduling and reduces peak load by 30%.

[0170] Standardized docking:

[0171] The solution complies with the ISO 50001 energy management system standard, and the data format is compatible with IEC 61970 (Energy Management Data Exchange). Indicators (such as entropy production rate) are mapped to ISO 50006 energy efficiency indicators, and the reporting format supports the verification requirements of ISO 50015.

[0172] Provide API interface to connect with smart grid platforms (such as State Grid EMS) to transmit comprehensive scores and optimization suggestions in real time.

[0173] Effect: Scalability supports campuses ranging from 500 to 5,000 households. Standardized docking reduces integration costs by 20%, making it suitable for scenarios such as hospitals and campuses, expanding the scope of application.

[0174] Need to explain Figure 3In the simulation diagram, the top figure shows the seasonal variation trend of the comprehensive evaluation score. The horizontal axis is "season" (spring, summer, autumn, winter), and the vertical axis is "comprehensive evaluation score" (unit: dimensionless, 0-100). The scores are 86.16, 90.69, 88.43, and 95.22, respectively. The highest scores are in winter, indicating that the system has the best performance and adaptability in winter. Summer is second, and spring and autumn are slightly lower. This may be due to demand fluctuations affecting the optimization effect. The bottom figure shows the spatial distribution of resource allocation efficiency. The horizontal axis is "X (meters)" (0-10 meters), and the vertical axis is "Y (meters)" (0-10 meters). The heat map color shows the efficiency distribution from blue (low, 0.2) to red (high, 0.8). The efficiency is higher in the center area and lower in the upper left corner, reflecting spatial unevenness in resource allocation and the need for layout optimization.

[0175] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further restrictions, an element defined by the statement "comprising a reference structure" does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0176] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-dimensional quantitative evaluation method for the operation of a multi-season park integrated energy system, characterized by: The steps include: Collect multi-source data, including heat flow data, gas flow data, water flow data, electricity flow data, inter-device electricity flow data, spatial location data, energy usage behavior data, questionnaire data, and historical operation strategy data; preprocessing the collected multi-source data, and constructing a dissipative structure model, an ecological network model, a group behavior model, a spatiotemporal trajectory model, a thermodynamic constraint model, and a cultural evolution model based on the preprocessed multi-source data; Based on the constructed model, quantitative indicators of each dimension are generated to form a six-dimensional evaluation system of self-organization characteristics, device collaboration, user behavior impact, spatiotemporal coupling characteristics, theoretical performance limits and operation strategy adaptability; Based on the generated quantitative indicators of each dimension, a comprehensive analysis of the indicator data is performed; The indicator data and model output data are processed to generate visual data.

2. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: In the multi-source data collection step, the heat flow, gas flow, water flow, electricity flow, electricity flow between devices, spatial location and energy usage behavior data are collected and transmitted at a frequency of 1 Hz, the questionnaire data are collected and transmitted once a week, and the historical operation strategy data are collected and transmitted once a day.

3. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: In the multi-source data preprocessing and model building steps, for heat flow, gas flow, and electricity flow data, sliding window denoising and seasonal trend decomposition are used to fill missing values ​​and generate standardized time series data for constructing a dissipative structure model. Nonlinear dynamic equations are used to describe system evolution and output the system's entropy production rate and self-organization state. For the power flow, heat flow, and gas flow data between devices, the device ID is mapped to a network node, and the energy flow is mapped to a directed edge. After normalization, an ecological network model is constructed to analyze the energy interaction efficiency between devices. For user energy usage behavior data and questionnaire data, the module uses K-means clustering to group user behavior and conducts time series correlation analysis to build a group behavior model to reflect user behavior patterns; coordinates of spatial location data are standardized, time windows are segmented based on energy flow time distribution data, and a space-time trajectory model is constructed to analyze resource transfer efficiency; energy balance calibration and entropy production distribution calculation are performed on heat flow, gas flow, and electricity flow data, a thermodynamic constraint model is constructed, and performance limits are evaluated; entropy value calculation of strategy parameters and time series analysis are performed on historical operation strategy data, a cultural evolution model is constructed, and strategy adaptability is analyzed.

4. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: In the quantitative indicator generation step, the entropy production rate and self-organization efficiency are calculated based on the time series analysis of the ARIMA model and seasonal difference processing; the dissipation gradient index is further calculated by the slope of the entropy production rate change in different seasons; the reciprocity index is calculated based on the graph theory algorithm of PageRank, and the circulation rate is calculated based on the closed-loop path analysis algorithm; the behavior consistency and system responsiveness index are calculated based on the statistical method of principal component analysis, and the impact of user behavior on the system is analyzed; the path analysis based on the Dijkstra algorithm is combined with weighted distance calculation to calculate the time and space accessibility and time compression rate; the energy balance analysis based on the Carnot cycle is used to calculate the limit efficiency ratio and loss concentration, and identify efficiency bottlenecks; the communication dynamics analysis based on the SIR model is combined with the dynamic adjustment of the strategy adoption rate to calculate the strategy diversity and communication efficiency.

5. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 4 is characterized in that: In the steps of calculating the entropy production rate, self-organization efficiency, and dissipation gradient, the standardized heat flow, gas flow, and electric energy flow time series are used as input, daily periodic fluctuations are removed, and an ARIMA (1,1,1) model with an autoregressive order of 1, a differencing order of 1, and a moving average order of 1 is configured. The sequence is fitted by maximum likelihood estimation, the entropy production rate is calculated based on the statistical mean of the fitted sequence, and the self-organization efficiency is the proportion of the stable part of the fitted sequence, ranging from 0 to 1; the dissipation gradient is calculated by the difference between the slopes of the fitted sequence in spring and winter; In the reciprocity index calculation step, the hourly data of electrical energy flow, thermal energy flow, and gas flow between devices are used as input to construct a 100×100 directed graph of the adjacency matrix, with the edge weight being the energy intensity, the PageRank value being initialized to 1 / 100, the damping factor being set to 0.85, and 10 iterations being performed. The reciprocity index is calculated as the normalized PageRank value in the range of 0-1. In the circulation rate calculation step, hourly energy flow data between devices is used as input to construct a directed graph, with edge weights being energy intensity. Closed-loop paths with more than three nodes are detected through depth-first search, and the circulation rate is calculated as the sum of the energy intensities of the closed-loop paths divided by the total edge energy intensity, with a range of 0-1.

6. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 4 is characterized in that: In the behavioral consistency calculation step, the daily electricity consumption data of 1,000 users is used as input to construct a 1,000×24 matrix, and the covariance matrix is ​​calculated after standardization. The first three principal components are extracted, and the behavioral consistency is the sum of the first three eigenvalues ​​divided by the sum of the total eigenvalues, ranging from 0 to 1; In the system responsiveness index calculation step, user behavior fluctuations and system operating parameters are used as input to construct a regression data set, fit a linear regression model, and estimate the regression coefficient. The system responsiveness is the mean of the regression coefficient. In the spatiotemporal reachability calculation step, the spatial position and energy flow time distribution data are used as input to construct a weighted graph with an edge weight of 0.5×distance+0.5×delay. The shortest path from the starting point to each node is calculated, and the spatiotemporal reachability is the average path length. In the time compression rate calculation step, the energy flow time distribution data is used as input, the low load period is identified through cluster analysis, and the redundant time period is compressed. The time compression rate is the ratio of the compressed time length, ranging from 0 to 1.

7. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 4 is characterized in that: In the step of calculating the limiting efficiency ratio and loss concentration, the hourly heat flow, gas flow, and electric energy flow data are used as input to calibrate the energy balance, calculate the actual efficiency and Carnot efficiency, the limiting efficiency ratio is the ratio of the actual efficiency to the Carnot efficiency, and the loss concentration is the spatial standard deviation of entropy production; In the strategy diversity and propagation efficiency calculation steps, 100 historical operation strategy data are used as input, the Shannon entropy of the parameters is calculated, the adoption rate time series is constructed, and an SIR model with an infection rate of 0.1 and a recovery rate of 0.05 is configured. After 100 iterations, the strategy diversity is calculated as the parameter entropy value, and the propagation efficiency is the final proportion of adopted strategies.

8. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: In the comprehensive analysis step, the multi-objective decision analysis algorithm based on TOPSIS is combined with the entropy method and expert scoring to determine the weight of each indicator, and the indicator data is weighted and normalized to generate a comprehensive evaluation score.

9. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: In the visualization data generation step, the indicator data and the model output data are transmitted at the same frequency, and time series interpolation and spatial gridding are performed on the indicator data and the model output data to generate visualization data. The seasonal variation trend of the indicator is displayed through a D3.js-based chart, and the spatial distribution of resource allocation efficiency is displayed through a heat map based on spatial interpolation. A visualization report containing various evaluation indicators is generated.

10. The multi-dimensional operation quantitative evaluation method of the multi-season park integrated energy system according to claim 1 is characterized in that: It also includes data cross-validation steps, analysis based on the Pearson correlation coefficient, comparison of data generated from different evaluation directions, calculation of correlation matrix and significance test, and generation of a consistency analysis report for evaluation results.

Images