Energy-saving effect evaluation method and system for perlite composite board based on edge computing

Abstract

The application discloses a perlite composite board energy-saving effect evaluation method and system based on edge computing, and particularly relates to the field of composite board energy-saving evaluation, and comprises the following steps: collecting thermal engineering data through a sensor, then inputting a heat transfer analysis model to identify thermal engineering defects and trigger the generation of an energy-saving evaluation report, and dynamically updating the model based on performance attenuation to realize closed-loop monitoring and evaluation of the performance of the perlite composite board. The perlite composite board energy-saving effect evaluation method and system based on edge computing realizes the automatic association of evaluation data with project progress plans and cost accounts by adding business metadata, improves business collaboration efficiency, realizes the instantaneous identification and accurate positioning of local thermal engineering defects by means of a lightweight heat transfer analysis model, provides complete decision-making basis for fine management, and realizes the automatic linkage of thermal engineering defect information and cost data by querying a cost mapping table of an edge computing node based on the first energy-saving index, and further improves business collaboration efficiency.

Description

Technical Field

[0001] This invention relates to the field of energy-saving assessment technology for composite panels, and more specifically, to a method and system for assessing the energy-saving effect of perlite composite panels based on edge computing. Background Technology

[0002] With the rapid development of green building and high-performance building materials, the construction industry's demand for energy-saving performance evaluation of perlite composite boards has shifted from parameter measurement to serving as a basis for business decision-making to support the entire construction lifecycle management. Traditional technology mainly involves measuring the steady-state thermal conductivity of perlite composite boards through a standard test chamber, inputting the data into energy consumption simulation software, and relying on a centralized cloud server to complete the annual energy consumption calculation and obtain an energy-saving rating. However, actual measurements show that temperature and humidity fluctuations, thermal bridging effects, and construction errors in the real building environment can cause deviations between the theoretical energy-saving rating and the actual energy consumption cost, leading to cost overruns, repeated adjustments to construction plans, and impacting the overall project progress. Furthermore, the long-term flow of evaluation data makes it impossible to synchronize it to the construction management end in a timely manner. When instantaneous thermal defects occur, it is impossible to quickly respond and adjust the procedures, resulting in the accumulation of quality problems and extended acceptance cycles.

[0003] To address the aforementioned issues, existing technologies deploy sensors at key nodes on each floor and utilize the Internet of Things (IoT) to upload data to a cloud server for centralized analysis. This initially shifts the computational focus from a remote data center to the construction site, achieving millisecond-level real-time sampling and second-level rapid calculation. This shortens the feedback cycle of assessment data and, to a certain extent, reflects the actual operating conditions of the building. It lays the foundation for energy-saving assessments to move from purely theoretical simulations to data-driven approaches, thereby improving the objectivity of the assessments.

[0004] However, in practical use, it still has some shortcomings. For example, it still relies on a centralized cloud for in-depth analysis and archiving. The evaluation results need to be manually entered again before they can be linked with the project schedule and cost ledger, resulting in low efficiency of business collaboration. In addition, the inherent inefficiency and discontinuity of the phased manual inspection as the main data supplementation method have led to the fragmentation of the management process and the incompleteness of the decision-making basis. This makes it difficult to achieve instantaneous identification and accurate location of local thermal defects, and cannot meet the needs of modern building construction refined management. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, the present invention provides a method and system for evaluating the energy-saving effect of perlite composite panels based on edge computing, which solves the problems mentioned in the background art through the following solutions.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] An edge computing-based method for evaluating the energy-saving performance of perlite composite panels includes:

[0008] S1: By deploying a sensor network at the perlite composite board components and their connections, on-site temperature and heat flux density data are continuously collected at a preset frequency to generate the first thermal state for energy consumption assessment.

[0009] S2: Edge computing nodes deployed at the construction site receive the first thermal state, perform data preprocessing, and generate a standardized second thermal state;

[0010] S3: Input the second thermal state into the lightweight heat transfer analysis model to identify and output the first energy-saving index containing the location of thermal defects;

[0011] S4: Based on a preset early warning rule, the first energy-saving indicator is judged:

[0012] If the warning conditions are met, an energy-saving optimization suggestion signal containing the location of the thermal defect will be triggered immediately;

[0013] S5: Continuously collect the third set of thermal parameters and execute:

[0014] Based on the energy-saving optimization suggestion signal, an energy-saving effect evaluation report is generated, which includes at least the energy-saving effect trend and the energy-saving effect ranking.

[0015] The thermal insulation performance of the perlite composite board is predicted to decrease, and the parameters of the lightweight heat transfer analysis model are dynamically updated.

[0016] Preferably, step S1, generating the first thermal state, specifically includes:

[0017] Each sensor node in the sensor network adds service metadata before sending out the collected temperature and heat flux density data;

[0018] The business metadata includes at least the project schedule task ID and cost ledger material batch number corresponding to the installation location of each sensor node.

[0019] Preferably, in S3, the lightweight heat transfer analysis model is a three-layer fully connected neural network;

[0020] Its input layer receives the second thermal state of S2, the hidden layer uses the ReLU activation function to perform feature nonlinear mapping, and the output layer uses the linear activation function to output the performance deviation rate between the actual thermal resistance value and the theoretical thermal resistance value at each sensor node location.

[0021] Preferably, in S3, the first energy-saving indicator includes at least the component spatial coordinates, actual thermal resistance value, performance deviation rate, defect area, defect type, and a predefined business action code.

[0022] Preferably, S3, the identification of defect type and area in the first energy-saving index, specifically includes:

[0023] Based on each sensor node in the second thermal state spatial coordinates ,temperature and heat flux density amplitude The two-dimensional temperature field on the surface of perlite composite board was reconstructed using the inverse distance weighting method. With heat flux density amplitude ;

[0024] Based on Fourier's law of heat conduction, the heat flux vector field is calculated and generated. ;

[0025] In the heat flow vector field In the process, several virtual probes are initialized based on predefined theoretical high-incidence areas of thermal bridging. Each virtual probe... Path tracing is performed based on iterative movement rules to form a heat flow path network;

[0026] Extract heat flow vectors from the heat flow path network. The continuous path serves as the thermal bridge framework, in which Represented as the entire scalar field of heat flux density The mean, It is represented as a dynamic threshold coefficient. At the same time, with the thermal bridge skeleton as the center line, it dynamically expands according to the local heat flux density to generate a thermal defect region and calculate its area.

[0027] The defect types are classified according to the topology of the thermal bridge skeleton and the theoretical connection method of the components in which it is located.

[0028] Preferably, in step S3, the heat flow path network is formed, and each virtual probe... The required iterative move rules include:

[0029] Each virtual probe At time step Location Its iterative movement rule is defined as:

[0030]

[0031] in, This is represented as the preset movement step size. Represented as a virtual probe The heat flow vector at the current location.

[0032] Preferably, S3 is a width that dynamically expands based on the local heat flux density, with the thermal bridge skeleton as the center line. The calculation formula is as follows:

[0033]

[0034] in, This is indicated by the position on the path of the thermal bridge skeleton. This is expressed as the heat flux density at that location. These represent the maximum and minimum heat flux densities on the thermal bridge frame, respectively. This represents the preset maximum and minimum expansion width.

[0035] Preferably, step S5 generates an energy-saving effect evaluation report, including:

[0036] Based on the first energy-saving index, calculate the dynamic change rate of at least one key parameter to generate a visual performance trend chart.

[0037] The energy-saving effects of multiple perlite composite board components were quantitatively scored and prioritized.

[0038] To achieve the above objectives, the present invention provides the following technical solution: an edge computing-based perlite composite board energy-saving effect evaluation system, which implements the above-mentioned edge computing-based perlite composite board energy-saving effect evaluation method, including:

[0039] Real-time acquisition module: Used to continuously acquire on-site temperature and heat flux density data at a preset frequency through a sensor network deployed at the perlite composite board components and their connections, and generate the first thermal state for energy consumption assessment;

[0040] Edge preprocessing module: used by edge computing nodes deployed at the construction site to receive the first thermal state, perform data preprocessing, and generate a standardized second thermal state;

[0041] Heat transfer model analysis module: used to input the second thermal state into the lightweight heat transfer analysis model to identify and output the first energy-saving index containing the location of thermal defects;

[0042] Real-time alarm module: used to judge the first energy-saving index based on a preset early warning rule.

[0043] If the warning conditions are met, an energy-saving optimization suggestion signal containing the location of the thermal defect will be immediately triggered;

[0044] Energy-saving effect assessment and feedback module: used to continuously collect the third set of thermal parameters and execute:

[0045] Based on the energy-saving optimization suggestion signal, an energy-saving effect evaluation report is generated, which includes at least the energy-saving effect trend and the energy-saving effect ranking.

[0046] The thermal insulation performance of the perlite composite board is predicted to decrease, and the parameters of the lightweight heat transfer analysis model are dynamically updated.

[0047] Preferably, the real-time alarm module has preset warning rules, specifically including:

[0048] Based on the defect area, performance deviation rate and defect type in the first energy-saving index, a cost mapping table stored in the edge computing node is queried in real time to dynamically calculate and quantify a cost impact index.

[0049] The cost impact index is encapsulated in the energy-saving optimization suggestion signal.

[0050] The technical effects and advantages of this invention are as follows:

[0051] 1. This invention adds business metadata to the temperature and heat flux density data collected by the sensor network, thereby realizing the automatic association between the evaluation data and the project schedule and cost ledger. This avoids the problem in the prior art that the evaluation results need to be manually entered again to link with the business ledger, and improves the efficiency of business collaboration.

[0052] 2. This invention achieves instantaneous identification and precise location of local thermal defects through a lightweight heat transfer analysis model, solving the problem of difficulty in achieving instantaneous identification and precise location of local thermal defects, and providing a complete decision-making basis for refined management;

[0053] 3. This invention achieves automatic linkage between thermal defect information and cost data by querying the cost mapping table of edge computing nodes based on the first energy-saving index, eliminating the need for manual secondary processing and further improving business collaboration efficiency. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the steps of an edge computing-based method for evaluating the energy-saving effect of perlite composite panels according to an embodiment of this application.

[0055] Figure 2 This is a block diagram of a perlite composite board energy-saving effect evaluation system based on edge computing, provided in an embodiment of this application. Detailed Implementation

[0056] 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.

[0057] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0058] Hereinafter, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first," "second," and "third" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0059] As attached Figure 1 The edge computing-based energy-saving evaluation method for perlite composite panels, as shown, collects thermal data through sensors, inputs it into a heat transfer analysis model to identify thermal defects and trigger the generation of an energy-saving evaluation report, and dynamically updates the model based on performance degradation to achieve closed-loop monitoring and evaluation of the perlite composite panel performance. Specifically, it includes the following steps:

[0060] S1: By deploying a sensor network at the perlite composite board components and their connections, on-site temperature and heat flux density data are continuously collected at a preset frequency to generate the first thermal state for energy consumption assessment.

[0061] S2: Edge computing nodes deployed at the construction site receive the first thermal state, perform data preprocessing, and generate a standardized second thermal state;

[0062] S3: Input the second thermal state into the lightweight heat transfer analysis model to identify and output the first energy-saving index containing the location of thermal defects;

[0063] S4: Based on a preset early warning rule, the first energy-saving indicator is judged:

[0064] If the warning conditions are met, an energy-saving optimization suggestion signal containing the location of the thermal defect will be triggered immediately;

[0065] S5: Continuously collect the third set of thermal parameters and execute:

[0066] Based on the energy-saving optimization suggestion signal, an energy-saving effect evaluation report is generated, which includes at least the energy-saving effect trend and the energy-saving effect ranking.

[0067] The thermal insulation performance of the perlite composite board is predicted to decrease, and the parameters of the lightweight heat transfer analysis model are dynamically updated.

[0068] Specifically, in S1, the first thermal state includes at least the temperature, heat flux density, ambient wind speed, solar radiation intensity, and theoretical thermal conductivity of the corresponding components for each sensor node.

[0069] In this embodiment, to analyze the thermal bridge effect, digital MEMS temperature sensors are preferably deployed on the surface of the plate and at the joints to collect the ambient temperature and the surface temperature of the plate. Their performance parameters are: measurement range -40℃ to 125℃, accuracy ±0.2℃, and resolution 0.1℃. To directly measure the thermal insulation performance, thin-film heat flux sensors based on the thermopile principle are deployed in pairs at the same locations to monitor the heat flux density passing through the perlite composite plate. Their performance parameters are: range 0-2000W / m. 2 With an accuracy of ±3%FS and a response time of <1 second, an ultrasonic anemometer was deployed in the monitoring area to capture the wind speed around the building components. Its performance parameters are: range 0-40 m / s, accuracy ±0.1 m / s. To correct for the heat gain effect from solar radiation, a thermopile-type total radiation sensor was deployed to measure the solar radiation intensity acting on the building surface. Its performance parameters are: range 0-2000 W / m². 2 Accuracy ±5%.

[0070] Furthermore, the deployment density of the sensor network is dynamically adjusted according to the different thermal zones of the target building. A predefined rule base for key thermal nodes is established within the edge computing nodes. This rule base includes, but is not limited to, rules such as the completion of all exterior window installations and the completion of splicing of continuous exterior wall panels exceeding 20 meters. When the project progress status is updated and matches any node in the rule base, an electronic work order containing the target area coordinates and the required number of sensors is automatically generated. Construction personnel use the electronic work order to deploy a plug-and-play mobile sensor cluster. After powering on, the mobile sensor cluster automatically broadcasts a network access request. Upon receiving the request, the edge gateway assigns a temporary network address and issues a configuration command containing the target area coordinates. Each sensor node then performs self-localization and transmits its precise location information as spatial registration information, completing the construction of the high-density monitoring network.

[0071] It should be noted that the sensor node achieves self-positioning in at least the following ways: it has a built-in GPS module to directly read its own latitude and longitude coordinates; in an indoor environment without GPS signal, it calculates its own coordinates by receiving the signal strength of multiple edge gateways at known locations and using triangulation.

[0072] Furthermore, the sensor network employs a low-power wireless personal area network (LPPAN) protocol for self-organizing network communication; the LPPAN uses ZigBee PRO or LoRaWAN protocols, and the network adopts a mesh topology, allowing data to be routed to the edge gateway via multi-hop relays, and employs an adaptive routing algorithm based on a route discovery protocol to enhance coverage and robustness in complex environments; the preset frequency is a millisecond-level sampling frequency; and the deployment strategy is to use a higher deployment density at the component connections of the perlite composite board than at the center of the board surface.

[0073] It should be noted that the preset frequency is a dynamic value set based on the statistical characteristics of historical data of perlite composite boards under normal working conditions. In this embodiment, the edge computing node periodically analyzes the historical data and sends the frequency corresponding to twice the standard deviation of the average temperature change rate in the past 5 minutes to the sensor node for execution.

[0074] In one possible implementation, each sensor node in the sensor network attaches business metadata before sending out the collected temperature and heat flux density data; the business metadata includes at least the project schedule task ID and cost ledger material batch number corresponding to the installation location of each sensor node.

[0075] It should be noted that the appending of the business metadata is completed before or during the preprocessing of sensor data by edge computing nodes. By pre-building a mapping table between sensor locations and business information, the mapping table is configured in each edge computing node when the sensors are deployed. When each edge computing node receives data packets transmitted by each sensor node, it queries the mapping table based on the sensor ID in the data packet and automatically appends the corresponding project schedule task ID and cost ledger material batch number to it.

[0076] It should be noted that the data packet corresponding to the first thermal state includes at least: sensor ID, acquisition time, measured values ​​of each parameter, physical location of the sensor, project schedule task ID, and cost ledger material batch number.

[0077] Specifically, in S2, in order to realize the real-time processing of the first thermal state transmitted by S1 and to provide high-quality, standardized data for subsequent energy-saving assessment and management collaboration, namely the second thermal state, S2 adopts the method of acquiring data characteristics locally, completely avoiding dependence on centralized cloud, and generating structured information that can be directly driven.

[0078] Furthermore, the data preprocessing includes: creating two logically independent but physically contiguous circular buffers as a double-buffered queue in the memory of the edge computing node; in this embodiment, the capacity of each buffer is set to accommodate 10 seconds of millisecond-level sampled data; when buffer A is receiving the first thermal state from S1, the data cleaning unit simultaneously reads data from the already full buffer B for processing; when buffer A is full, a pointer switch is executed to write data into buffer B, while the data cleaning unit reads from buffer A.

[0079] Furthermore, the data preprocessing also includes: a data cleaning unit: to remove outliers from the data stream, a sliding window-based method is used. The principle, for a continuously input data stream, is that this embodiment uses each current data point... Construct a sliding window of fixed length N=500, and calculate the mean of all data points within the window. with standard deviation ;like Then determine Outliers are identified and marked and removed from the data stream, and replaced with the linear interpolation result of the two valid data points before and after them. All removed outliers, along with their original data, replacement values, and timestamps, are recorded in an anomaly register. This register records not only the outliers but also their corresponding sensor IDs and spatial coordinates, automatically associated with the component codes, providing direct evidence for subsequent defective component location. The noise reduction unit uses a first-order low-pass filter to filter out high-frequency noise in the data stream, implemented using an exponentially weighted moving average algorithm. Its smoothing coefficient is set to 0.1, determined by testing under typical noise conditions in the field to improve the signal-to-noise ratio while maintaining signal strength. The optimization goal is to achieve the best balance between phase lag and signal lag. After multiple tests and calibrations, the optimal balance was determined to effectively suppress random fluctuations while preserving the main trend of the data. Standardization unit: To eliminate the influence of different sensor dimensions, a minimum-maximum normalization method is adopted to linearly map the data of each channel to an interval and dynamically determine the normalization boundary value. In this embodiment, the boundary value is dynamically updated based on a sliding window with a length of 1 hour. By continuously maintaining the measured maximum and minimum values ​​within the window, the normalization calculation uses this dynamic boundary to adapt to the large changes in temperature and humidity in the construction environment during the day and night and seasonally, ensuring that the data input to S3 is always within a stable and comparable range.

[0080] It should be noted that the second thermal state maintains strict time alignment with the first thermal state by preserving the high-precision timestamps of each data point from the sensors, and all preprocessing operations do not change the temporal relationship of the data points, ensuring the accuracy of subsequent analysis in the time dimension. The generated second thermal state is transmitted to the lightweight heat transfer analysis model of S3 through a local shared memory interface. The real-time statistics table and anomaly register generated during the preprocessing process are used to collect data reception rate and outlier removal rate, which are automatically synchronized to the schedule plan and cost ledger in real time without any manual secondary entry.

[0081] Specifically, in S3, a lightweight heat transfer analysis model deployed on edge computing nodes enables real-time analysis of the energy-saving effect of perlite composite panels and precise location of thermal defects; thus constituting at least the component spatial coordinates, theoretical thermal resistance, actual thermal resistance, and performance deviation rate. The first energy-saving indicator includes the defect area, defect type, and a predefined business action code.

[0082] It should be noted that the predefined business action code has a mapping rule based on a conditional logic table with multiple fields such as defect type, defect area, performance deviation rate, and defect location. This table is pre-placed in the edge nodes, enabling technical indicators to be automatically and unambiguously mapped to specific project management instructions. This table is also pre-placed in the configuration files of each edge computing node. In this embodiment, the mapping rule includes, but is not limited to: if the defect type is a linear thermal bridge and the defect area is >2m². 2 Then the business action code is schedule adjustment; if the performance deviation rate If the defect type is a point defect and it is located around the door or window opening, the business action code is "immediate inspection".

[0083] Furthermore, the lightweight heat transfer analysis model is a three-layer fully connected neural network with 128, 64, and 1 neurons respectively. The input is the second thermal state, and the output is the actual thermal resistance value at the location of each sensor node. Compared with theoretical thermal resistance deviation rate The lightweight heat transfer analysis model is trained under supervised conditions using historical data and corresponding laboratory standard measurements. The loss function is mean squared error loss, the hidden layer activation function is ReLU, and the output layer activation function is linear activation. More than 15% of the sensor nodes were selected as candidate defect points.

[0084] It should be noted that the historical data includes second thermal state samples collected from multiple deployed perlite composite board buildings under different seasons and weather conditions, after S2 pretreatment; the laboratory standard measurement value is the baseline thermal resistance value measured on the same batch of perlite composite board samples using the protective hot plate method under constant temperature and humidity conditions in the laboratory.

[0085] In one possible implementation, the defect area and type in the first energy-saving index can be obtained based on the candidate defect points using the following methods:

[0086] For edge nodes with extremely limited computing resources, a density-based spatial clustering algorithm is used to directly divide the candidate defect points into defect regions. Each cluster is identified as a defect region, and its convex hull area is calculated. The defect type is initially determined based on the aspect ratio of the cluster shape. Clusters with an aspect ratio greater than 3 are classified as linear defects, otherwise they are classified as planar defects. In one embodiment, the neighborhood radius of the spatial clustering algorithm is set to 1.5 times the average spacing between sensors, and the minimum number of points is set to 5.

[0087] Based on each sensor node in the second thermal state spatial coordinates ,temperature and heat flux density amplitude The two-dimensional temperature field on the surface of the perlite composite board was reconstructed using the inverse distance weighting method. With heat flux density scalar field The power parameter of the inverse distance weighting method is set to 2; based on Fourier's law of heat conduction, the heat flux vector field is calculated and generated. ,in, This represents the thermal conductivity of the perlite composite board, and the value is the one specified in the factory inspection report for the target batch of perlite composite boards. Represented as temperature field The gradient; in the heat flux vector field In this process, several virtual probes are initialized in a predefined theoretical high-incidence area of ​​thermal bridges; in one embodiment, 10 virtual probes are randomly initialized within a circular area with a radius of 0.5 meters around each candidate defect point; each virtual probe At time step Location Its iterative movement rule is defined as:

[0088]

[0089] in, This represents the preset movement step size, which can be defined as 0.05m. Represented as a virtual probe The heat flow vector at the current location, The magnitude is represented as a vector; each virtual probe Path tracing is performed based on the iterative movement rules to form a heat flow path network. Path tracing terminates when the virtual probe moves to the boundary of the computational region, or when the Euclidean distance of its three consecutive iterations is less than 0.01 meters. From the heat flow path network, all path points that satisfy the following conditions are extracted. The continuous path serves as the thermal bridge framework, in which Represented as the entire scalar field of heat flux density The mean, This is represented as a dynamic threshold coefficient, with a value range of 1.3 to 1.8; simultaneously, with the thermal bridge skeleton as the centerline, the expansion width is dynamically calculated based on the local heat flux density. The calculation formula is as follows:

[0090]

[0091] in, This is indicated by the position on the path of the thermal bridge skeleton. This is expressed as the heat flux density at that location. These represent the maximum and minimum heat flux densities on the thermal bridge frame, respectively. The maximum and minimum expansion widths are represented by preset values, which are then used to generate a polygonal thermal defect region and calculate its area. The defect type is classified according to the topology of the thermal bridge skeleton and the theoretical connection method of the component it is located in. In one embodiment, if the thermal bridge skeleton is a continuous curve and its length is greater than 1 meter, it is classified as a linear thermal bridge; if it is a closed curve, it is classified as a planar defect; and in other cases, it is classified as a point defect.

[0092] Specifically, in S4, the first energy-saving index output by S3 is transformed into automated instructions that can guide on-site construction. Through early warning rules deployed on each edge computing node, the thermal defects are identified instantly, and energy-saving optimization suggestion signals linked to cost and schedule are automatically generated. The energy-saving optimization suggestion signals are transmitted through JSON format data packets, which contain at least the following fields: a unique identifier for the defect generated by the location coordinate hash, the early warning level, the spatial coordinates of the component, construction optimization suggestions, and the cost impact index.

[0093] It should be noted that the preset early warning rule includes a threshold table stored locally on the edge computing node, which defines the parameter thresholds corresponding to different early warning levels; the structure of the threshold table includes at least an early warning level, an area threshold, a performance deviation rate threshold, and a duration threshold; in one embodiment, when the performance deviation rate in the first energy-saving indicator exceeds 30% for a duration greater than 60 seconds, and the defect area is greater than 2.0 square meters, it is determined that the 'medium-level early warning' condition is met.

[0094] Specifically, in S5, the third set of thermal parameters is continuously collected, which is the data accumulated over a long period of time and preprocessed by S2, and macroscopic assessment and self-optimization are performed based on the energy-saving optimization suggestion signal from S4.

[0095] In one possible implementation, generating an energy-saving effect assessment report includes: calculating the dynamic change rate of at least one key parameter based on the actual thermal resistance value of the first energy-saving index to generate a visualized performance trend chart; the dynamic change rate is obtained by calculating the month-on-month growth rate of the moving average of the current period and the previous period; and quantifying and prioritizing the energy-saving effects of multiple perlite composite board components.

[0096] In one embodiment, each perlite composite panel component is quantitatively scored, and the quantitative score is... The formula is expressed as:

[0097]

[0098] in, This represents the actual thermal resistance value transferred by the S3 lightweight heat transfer model during the current evaluation period. This represents the theoretical thermal resistance value of perlite composite boards of the same batch and with the same connection method under factory-dried conditions. This is represented by the number of historical alarms within the most recent 10 consecutive evaluation periods, where 0.6 and 0.4 are empirical values; then, based on the scoring results, all components are ranked according to their energy-saving effects to form a priority list.

[0099] Furthermore, the prediction of the thermal insulation performance degradation of the perlite composite board is achieved by using a random forest regression algorithm to train and predict a continuously collected set of third thermal parameters; in one embodiment, the input features include: component year, history, etc. The mean and seasonal index, the forecast target is the next 30 days. The value is set to be retrained every 24 hours using the third set of thermal parameters.

[0100] Furthermore, dynamically updating the parameters of the lightweight heat transfer analysis model involves feeding back the knowledge learned by the prediction model to the model in S3 through the parameter write-back interface.

[0101] In one embodiment, when the prediction model achieves an average prediction accuracy of over 95% during a 15-day validation period, a dynamic update is triggered: the learned long-term decay trend is converted into a correction coefficient to adjust the numerical table of theoretical thermal resistance referenced by the lightweight heat transfer analysis model in S3; if the false alarm rate of S3 increases by more than 50% within 24 hours after the update, the rollback mechanism is automatically activated to restore the old parameters.

[0102] As attached Figure 2 The energy-saving effect evaluation system for perlite composite boards based on edge computing shown includes: a real-time acquisition module, an edge preprocessing module, a heat transfer model analysis module, a real-time alarm module, and an energy-saving effect evaluation and feedback module.

[0103] The real-time acquisition module is used to continuously acquire on-site temperature and heat flux density data at a preset frequency through a sensor network deployed on the perlite composite board components and their connections, and generate a first thermal state for energy consumption assessment.

[0104] The edge preprocessing module is used by edge computing nodes deployed at the construction site to receive the first thermal state, perform data preprocessing, and generate a standardized second thermal state.

[0105] The heat transfer model analysis module is used to input the second thermal state into the lightweight heat transfer analysis model to identify and output a first energy-saving index that includes the location of thermal defects.

[0106] The real-time alarm module is used to judge the first energy-saving index based on a preset early warning rule: if the early warning conditions are met, an energy-saving optimization suggestion signal containing the location of the thermal defect is immediately triggered.

[0107] In one possible implementation, the preset early warning rule includes: based on the defect area, performance deviation rate and defect type in the first energy-saving index, querying a cost mapping table stored in the edge computing node in real time to dynamically calculate and quantify a cost impact index; the cost impact index is encapsulated in the energy-saving optimization suggestion signal; and the cost impact index is automatically and in real time synchronized to the cost ledger module of the project management system through the communication interface as a budget change item to be confirmed.

[0108] It should be noted that the cost mapping table includes at least defect type codes, estimated working hours, and cost index calculation formulas. In one embodiment, the cost index calculation formula can be expressed as the ratio of the product of estimated working hours and unit working hour cost to the total project budget. The communication interface is a publish / subscribe interface based on the MQTT protocol. Its edge computing node acts as the publisher, publishing the encapsulated energy-saving optimization suggestion signal and listening to the cost ledger unit. Once a message is received, a budget change entry with a status of 'pending confirmation' is created in the cost ledger, thereby achieving real-time synchronization without the need for manual secondary entry.

[0109] The energy-saving effect evaluation and feedback module is used to continuously collect the third set of thermal parameters and perform the following: based on the energy-saving optimization suggestion signal, generate an energy-saving effect evaluation report, which includes at least the energy-saving effect trend and the energy-saving effect ranking; predict the attenuation of the thermal insulation performance of the perlite composite board and dynamically update the parameters of the lightweight heat transfer analysis model.

[0110] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0111] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the energy-saving effect of perlite composite panels based on edge computing, characterized in that, include: S1: By deploying a sensor network on the perlite composite board components and their connections, on-site temperature and heat flux density data are continuously collected at a preset frequency to generate a first thermal state for energy consumption assessment; wherein, the sensor network adopts a low-power wireless personal area network protocol for self-organizing network communication, and the deployment density is dynamically adjusted according to the thermal zoning of the target building. The first thermal state includes: each sensor node in the sensor network attaches business metadata before sending out the collected temperature and heat flux density data; the business metadata includes at least the project schedule task ID and cost ledger material batch number corresponding to the installation location of each sensor node; S2: Edge computing nodes deployed at the construction site receive the first thermal state and perform data preprocessing locally to generate a standardized second thermal state; the data preprocessing includes: using a double-buffered circular queue mechanism to realize pipeline-style cleaning of the data stream, and synchronously generating a real-time statistical table and an anomaly register; the real-time statistical table and the anomaly register are automatically synchronized to the cost ledger module of the project management system; S3: The second thermal state input is deployed on the lightweight heat transfer analysis model on the edge computing node, so that the edge computing node can identify and output a first energy-saving index containing the location of thermal defects locally; the lightweight heat transfer analysis model is a neural network model, and its output is the performance deviation rate at the location of each sensor node; S4: The first energy-saving index is judged based on a preset early warning rule stored in the edge computing node: If the warning conditions are met, the edge computing node will immediately trigger an energy-saving optimization suggestion signal containing the location of the thermal defect and the cost impact index, and automatically synchronize the cost impact index to the cost ledger module through the communication interface. S5: The edge computing node continuously collects the third set of thermal parameters and executes the following on the edge computing node: Based on the energy-saving optimization suggestion signal, an energy-saving effect evaluation report is generated, which includes at least the energy-saving effect trend and the energy-saving effect ranking. Furthermore, the thermal insulation performance of the perlite composite board is predicted to decrease, and the parameters of the lightweight heat transfer analysis model deployed on the edge computing node are dynamically updated based on the prediction results.

2. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 1, characterized in that: The S3 lightweight heat transfer analysis model is a three-layer fully connected neural network. Its input layer receives the second thermal state of S2, the hidden layer uses the ReLU activation function to perform feature nonlinear mapping, and the output layer uses the linear activation function to output the performance deviation rate between the actual thermal resistance value and the theoretical thermal resistance value at each sensor node location.

3. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 1, characterized in that: The first energy-saving indicator, S3, includes at least the component spatial coordinates, actual thermal resistance value, performance deviation rate, defect area, defect type, and a predefined business action code.

4. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 3, characterized in that: S3, the identification of defect types and areas in the first energy-saving index, specifically includes: Based on each sensor node in the second thermal state spatial coordinates ,temperature and heat flux density amplitude The two-dimensional temperature field on the surface of perlite composite board was reconstructed using the inverse distance weighting method. With heat flux density amplitude ; Based on Fourier's law of heat conduction, the heat flux vector field is calculated and generated. ; In the heat flow vector field In the process, several virtual probes are initialized based on predefined theoretical high-incidence areas of thermal bridging. Each virtual probe... Path tracing is performed based on iterative movement rules to form a heat flow path network; Extract heat flow vectors from the heat flow path network. The continuous path serves as the thermal bridge framework, in which Represented as the entire scalar field of heat flux density The mean, It is represented as a dynamic threshold coefficient. At the same time, with the thermal bridge skeleton as the center line, it dynamically expands according to the local heat flux density to generate a thermal defect region and calculate its area. The defect types are classified according to the topology of the thermal bridge skeleton and the theoretical connection method of the components in which it is located.

5. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 4, characterized in that: In step S3, the heat flow path network is formed, and each virtual probe... The iterative movement rules to be executed include: Each virtual probe At time step Location Its iterative movement rule is defined as: ， in, This is represented as the preset movement step size. Represented as a virtual probe The heat flow vector at the current location.

6. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 4, characterized in that: S3 refers to the width that dynamically expands based on the local heat flux density, with the thermal bridge framework as the center line. The calculation formula is as follows: ， in, This is indicated by its position on the path of the thermal bridge skeleton. This is expressed as the heat flux density at that location. These represent the maximum and minimum heat flux densities on the thermal bridge frame, respectively. This represents the preset maximum and minimum expansion width.

7. The method for evaluating the energy-saving effect of perlite composite panels based on edge computing according to claim 1, characterized in that: Step S5 generates an energy-saving effect assessment report, including: Based on the first energy-saving index, calculate the dynamic change rate of at least one key parameter to generate a visual performance trend chart. The energy-saving effects of multiple perlite composite board components were quantitatively scored and prioritized.

8. An edge computing-based perlite composite board energy-saving effect evaluation system, comprising the edge computing-based perlite composite board energy-saving effect evaluation method according to any one of claims 1-7, characterized in that, include: Real-time acquisition module: Used to continuously acquire on-site temperature and heat flux density data at a preset frequency through a sensor network deployed at the perlite composite board components and their connections, and generate the first thermal state for energy consumption assessment; Edge preprocessing module: used by edge computing nodes deployed at the construction site to receive the first thermal state, perform data preprocessing, and generate a standardized second thermal state; Heat transfer model analysis module: used to input the second thermal state into the lightweight heat transfer analysis model to identify and output the first energy-saving index containing the location of thermal defects; Real-time alarm module: used to judge the first energy-saving index based on a preset early warning rule. If the warning conditions are met, an energy-saving optimization suggestion signal containing the location of the thermal defect will be triggered immediately; Energy-saving effect assessment and feedback module: used to continuously collect the third set of thermal parameters and execute: Based on the energy-saving optimization suggestion signal, an energy-saving effect evaluation report is generated, which includes at least the energy-saving effect trend and the energy-saving effect ranking. The thermal insulation performance of the perlite composite board is predicted to decrease, and the parameters of the lightweight heat transfer analysis model are dynamically updated.

9. The energy-saving effect evaluation system for perlite composite panels based on edge computing according to claim 8, characterized in that: The real-time alarm module has preset warning rules, specifically including: Based on the defect area, performance deviation rate and defect type in the first energy-saving index, a cost mapping table stored in the edge computing node is queried in real time to dynamically calculate and quantify a cost impact index. The cost impact index is encapsulated in the energy-saving optimization suggestion signal.

Images