Building operation and maintenance management method and system based on big data

Abstract

The invention discloses a building operation and maintenance management method and system based on big data, and relates to the technical field of building intellectualization, and the method comprises the steps: collecting multi-source heterogeneous data, executing data standardization processing, and generating a standard data stream; constructing a virtual building model based on a digital twinning technology, mapping the standard data stream to a corresponding device node of the virtual building model, and generating a device feature matrix; inputting the equipment characteristic matrix into a deep learning network, training to generate an energy consumption prediction model, and outputting an energy consumption prediction sequence; a multilayer perceptron health scoring model is constructed based on the energy consumption prediction sequence, and a fault early warning priority list is generated in combination with an equipment weight coefficient; and automatically generating a maintenance work order according to the fault early warning priority list. Through the maintenance effect evaluation and secondary maintenance mechanism, a complete closed-loop management process is formed, the maintenance quality is ensured, the maintenance strategy is continuously optimized, and the overall operation and maintenance management level is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent building technology, and in particular to a building operation and maintenance management method and system based on big data. Background Art

[0002] With the rapid development of intelligent buildings, the intelligence and automation level of building operation and maintenance management has been continuously improved. At present, most intelligent buildings have installed various sensors and automatic control systems, which can realize basic environmental monitoring, energy metering and equipment control functions. However, these systems often operate independently and lack a unified data management and analysis platform, resulting in the inability to fully utilize data resources and low operation and maintenance management efficiency.

[0003] Traditional building operation and maintenance management mainly relies on manual inspections and regular maintenance, which has the following problems: it is difficult to grasp the operating status of equipment in real time, and the fault warning capability is insufficient; the maintenance plan lacks pertinence and it is impossible to formulate a reasonable maintenance strategy based on the actual status of the equipment; the scheduling efficiency of maintenance resources is low and the response time is long; the maintenance record management is not standardized and the data of the entire life cycle of the equipment is difficult to trace. These problems have seriously restricted the improvement of the level of building operation and maintenance management. Summary of the invention

[0004] In view of the problems of existing building operation and maintenance management systems such as decentralized data collection, insufficient prediction capabilities, and low maintenance efficiency, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to realize the unified collection and analysis of building equipment operation data through the Internet of Things and big data technology, establish equipment health status assessment and fault warning mechanism, improve the efficiency of maintenance resource scheduling, and thus enhance the intelligent level of building operation and maintenance management.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, an embodiment of the present invention provides a building operation and maintenance management method based on big data, which includes collecting multi-source heterogeneous data through an Internet of Things sensor network, performing data standardization processing, and generating a standard data stream; constructing a virtual building model based on digital twin technology, mapping the standard data stream to the corresponding device nodes of the virtual building model, and using the spatiotemporal attention mechanism to extract the device operation characteristics to generate a device feature matrix; inputting the device feature matrix into a deep learning network, training and generating an energy consumption prediction model, and outputting an energy consumption prediction sequence; constructing a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generating a fault warning priority list in combination with the device weight coefficient; automatically generating a maintenance work order according to the fault warning priority list, allocating the work order based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, pushing it to the operation and maintenance personnel, and feeding back the work order execution results to the virtual building model for status update.

[0007] As a preferred solution of the building operation and maintenance management method based on big data described in the present invention, wherein: constructing a virtual building model based on digital twin technology includes the following steps: acquiring the building information model data of the building, constructing a three-dimensional geometric model of the building through a three-dimensional modeling engine, and generating a virtual building skeleton model; based on the virtual building skeleton model, calculating and constructing a spatial position relationship matrix; constructing a spatial attention network based on the spatial position relationship matrix, and generating spatial association weights between devices; The standard data stream is organized in time series, and the dynamic time series features of the device operation are extracted through the temporal attention network; the spatial correlation weights and dynamic time series features are fused to generate a spatiotemporal fusion feature vector; the spatiotemporal fusion feature vector is organized into a device feature matrix.

[0008] As an optimal solution of the building operation and maintenance management method based on big data described in the present invention, the training and generation of the energy consumption prediction model includes the following steps: slidingly dividing the equipment feature matrix according to a preset time window to generate a sample set, and dividing the sample set into a training sample set and a verification sample set according to a preset ratio; constructing a bidirectional long short-term memory network as the backbone network of the energy consumption prediction model, inputting the training sample set into the bidirectional long short-term memory network, and extracting the time series feature vector; setting a multi-head attention calculation unit in the output layer of the bidirectional long short-term memory network to generate a fused feature vector; inputting the fused feature vector into the fully connected layer to obtain the energy consumption prediction value, and calculating the mean square error loss value between the energy consumption prediction value and the actual energy consumption value; according to the calculated mean square error loss value, using an adaptive learning rate optimization algorithm to update the network parameters; inputting the verification sample set into the bidirectional long short-term memory neural network, calculating the verification loss value, and using an early stopping mechanism based on the convergence of the verification loss to select the optimal model parameters to obtain the energy consumption prediction model; inputting the real-time equipment feature matrix into the energy consumption prediction model, and outputting the energy consumption prediction sequence.

[0009] As a preferred solution of the building operation and maintenance management method based on big data described in the present invention, generating a fault warning priority list includes the following steps: using a differential calculation unit to perform a time series differential operation on the energy consumption prediction sequence to obtain an energy consumption change trend array; calling a time series pattern recognition module to perform feature extraction on the energy consumption change trend array to obtain fluctuation characteristic values, trend slope values ​​and maximum energy consumption difference values; inputting the fluctuation characteristic values, trend slope values ​​and maximum energy consumption difference values ​​into a pre-trained multi-layer perceptron to output a device health risk score; reading the device weight coefficient in the device configuration database, multiplying the device weight coefficient by the device health risk score to obtain the device warning priority score; executing a descending sorting algorithm to sort the device warning priority scores to generate an equipment fault warning priority list.

[0010] As a preferred solution of the building operation and maintenance management method based on big data described in the present invention, wherein: automatically generating a maintenance work order according to the fault warning priority list includes: calling the maintenance knowledge base, matching the corresponding standard maintenance process based on the equipment number and the warning level, and generating an initial maintenance work order; constructing an equipment maintenance difficulty assessment model, and calculating the equipment maintenance difficulty coefficient according to the equipment type, fault complexity, and maintenance man-hours; establishing an operation and maintenance personnel skill evaluation system, and quantifying the skill level based on the operation and maintenance personnel's work experience, professional qualifications, and historical maintenance success rate; querying the operation and maintenance personnel resource pool database, matching the equipment maintenance difficulty coefficient with the skill level, and adopting Use the task allocation algorithm to select operation and maintenance personnel who meet the qualification requirements and establish a work order allocation table; push the initial maintenance work order to the operation and maintenance personnel's mobile terminal through the operation and maintenance management platform, trigger the work order to start the execution process, and record the work order receipt time; receive the work order execution data sent back by the operation and maintenance personnel through the mobile terminal, and update the status information of the corresponding equipment node in the virtual building model based on the work order execution data; monitor the operating status of the equipment after maintenance, collect the performance data of the first operating cycle after maintenance, compare and analyze it with the performance data before maintenance, and generate a maintenance effect evaluation report; based on the results of the maintenance effect evaluation report, determine whether it is necessary to start a secondary maintenance process.

[0011] As a preferred solution of the building operation and maintenance management method based on big data described in the present invention, obtaining the energy consumption change trend array includes the following steps: constructing a time window matrix, and segmenting the energy consumption prediction sequence according to a preset window length N; performing a first-order difference operation on the time window matrix, calculating the difference between adjacent predicted energy consumption values ​​in each time window, and generating an initial difference array; performing a median filter on the initial difference array to generate a smoothed difference array; multiplying the smoothed difference array by a time step correction coefficient to obtain an energy consumption change rate array per unit time; using a wavelet transform on the energy consumption change rate array to extract the main trend components of the energy consumption change, and obtaining an energy consumption change trend array.

[0012] As a preferred solution of the building operation and maintenance management method based on big data described in the present invention, the construction process of the virtual building skeleton model includes: importing building information model data, the building information model data includes floor plans, equipment layout diagrams, and pipeline distribution diagrams; calling a geometric conversion engine to parse the building information model data and output a standard geometric description language file; parsing the standard geometric description language file, extracting building component features, and establishing a component feature table; based on the standard geometric description language file, performing a meshing operation to generate a three-dimensional mesh data structure; loading the three-dimensional mesh data structure and the building component feature table, calling a hierarchical reconstruction algorithm to assemble spatial components, and generating a virtual building skeleton model.

[0013] In the second aspect, an embodiment of the present invention provides a building operation and maintenance management system based on big data, which includes a data acquisition module for collecting multi-source heterogeneous data through an Internet of Things sensor network, performing data standardization processing, and generating a standard data stream; a mapping module for building a virtual building model based on digital twin technology, mapping the standard data stream to the corresponding device nodes of the virtual building model, using the spatiotemporal attention mechanism to extract the device operation characteristics, and generating a device feature matrix; a prediction and analysis module for inputting the device feature matrix into a deep learning network, training and generating an energy consumption prediction model, and outputting an energy consumption prediction sequence; an evaluation and early warning module for building a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generating a fault warning priority list in combination with the equipment weight coefficient; a work order management module for automatically generating a maintenance work order according to the fault warning priority list, allocating the work order based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, pushing it to the operation and maintenance personnel, and feeding back the work order execution results to the virtual building model for status update.

[0014] The beneficial effects of the present invention are as follows: the present invention realizes the real-time mapping of physical buildings and virtual models through the combination of the Internet of Things sensor network and the digital twin technology, provides an intuitive three-dimensional visualization interface, and greatly improves the efficiency of equipment management and fault location. The deep learning architecture that adopts a bidirectional long short-term memory network combined with a multi-head attention mechanism can accurately capture the timing characteristics of equipment operation data, and significantly improve the accuracy and prediction time span of energy consumption prediction. The fault warning mechanism based on equipment health scores and weight coefficients, combined with the intelligent work order distribution system, realizes timely detection and rapid response of faults, and improves the initiative and pertinence of operation and maintenance work. Through maintenance effect evaluation and secondary maintenance mechanism, a complete closed-loop management process is formed to ensure maintenance quality and continuously optimize maintenance strategies, thereby improving the overall operation and maintenance management level. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 This is a framework flow chart of the building operation and maintenance management method based on big data.

[0017] Figure 2 Obtain a flow chart for the equipment feature matrix of the big data-based building operation and maintenance management approach.

[0018] Figure 3 Obtain a flow chart for the fault warning priority list of the big data-based building operation and maintenance management method. DETAILED DESCRIPTION

[0019] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0021] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0022] Example 1, reference Figure 1 to Figure 3 , which is the first embodiment of the present invention, and this embodiment provides a building operation and maintenance management method based on big data. The framework flow chart is as follows Figure 1 As shown, including, S1: Collect multi-source heterogeneous data through the IoT sensor network, perform data standardization processing, and generate standard data streams.

[0023] Specifically, configure an IoT sensor network, which includes environmental monitoring sensors, energy monitoring equipment and occupancy sensors, wherein the environmental monitoring sensors include temperature sensors, humidity sensors and air quality sensors, the energy monitoring equipment includes smart electricity meters and smart water meters, and the occupancy sensors include crowd density sensors, space usage status sensors and equipment operation status sensors; collect multi-source heterogeneous data of the IoT sensor network through edge devices, the multi-source heterogeneous data include temperature and humidity data, air quality index data, energy consumption data, equipment status data and space occupancy data, and add a unified timestamp to the collected multi-source heterogeneous data for data annotation; perform format conversion on the multi-source heterogeneous data with a unified timestamp, such as converting BIM data into IFC format and converting energy data into B2MML format, to generate format standardized data; integrate and process the format standardized data, reorganize according to data type and timing relationship, and generate a standard data stream.

[0024] Preferably, by adopting a layered IoT sensor network, data collection efficiency is improved and system stability is enhanced. The data preprocessing function of edge devices effectively reduces the data transmission load. The addition of a unified timestamp solves the problem of inconsistent timing of data from different sources.

[0025] S2: Build a virtual building model based on digital twin technology, map the standard data stream to the corresponding device nodes of the virtual building model, use the spatiotemporal attention mechanism to extract the dynamic characteristics of the equipment operation status, and generate the equipment feature matrix.

[0026] Specifically, the device feature matrix acquisition flow chart is as follows: Figure 2 As shown, the following steps are included: S2.1: Obtain the building information model data of the building, construct a three-dimensional geometric model of the building through a three-dimensional modeling engine, and generate a virtual building skeleton model.

[0027] Specifically, the building information model data is imported, and the geometry conversion engine is called to perform format analysis on the building information model data, and a standard geometry description language file is output. The building information model data includes floor plans, equipment layout diagrams, and pipeline distribution diagrams; the standard geometry description language file is parsed, the geometric parameters and spatial information of wall components, door and window components, and stair components are extracted, and a component feature table is established. The component feature table records the spatial coordinates, shape parameters, and material properties of each component; the floor plan data in the standard geometry description language file is read, the meshing parameters are set, an adaptive meshing operation is performed, and a three-dimensional mesh data structure is generated; the three-dimensional mesh data structure and the building component feature table are loaded, and a hierarchical reconstruction algorithm is called to assemble the spatial components in a vertical layered order, and a virtual building skeleton model is output.

[0028] Furthermore, the spatial data of the equipment layout diagram is parsed, the spatial coordinates of the equipment installation position are calculated, the equipment nodes are marked in the virtual building skeleton model, and the equipment location information table is generated; the topological data of the pipeline distribution diagram is read, the spatial path of the pipeline is calculated, the pipeline direction is drawn in the virtual building skeleton model, and the pipeline connection topology table is established; the spatial interference detection algorithm is executed on the virtual building skeleton model, the equipment location information table and the pipeline connection topology table to eliminate the component collision points and output the optimized three-dimensional geometric model; the material attribute data in the building component feature table is extracted, the material mapping operation is performed, the material data is attached to the surface of the three-dimensional geometric model, and a visual model with material effects is generated; the geographic coordinate reference point of the physical building is measured, the coordinate transformation matrix is ​​calculated, the local coordinates of the visual model are mapped to the geographic coordinate system, and the calibrated virtual building skeleton model is output.

[0029] In specific implementation, the geometry conversion engine uses Open Cascade or VTK library to implement BIM format parsing and conversion. Adaptive meshing is based on the quadtree / octree algorithm, with the basic grid size set to 0.5m, the local encryption threshold set to 0.1m, and the grid quality factor set to no less than 0.85 to ensure grid quality. The hierarchical reconstruction algorithm uses a spatial octree structure to complete the assembly of walls, doors, windows, stairs, equipment, and pipelines in each layer in order from the bottom layer to the top layer. Spatial interference detection uses a hierarchical bounding box (AABB tree) for rapid detection, and performs precise triangular mesh interference detection on collision components. The coordinate transformation uses a seven-parameter transformation model, and the coordinates of the reference point are obtained through GPS measurement to ensure centimeter-level accuracy.

[0030] It is worth noting that this step uses an adaptive meshing algorithm, which can automatically adjust the mesh density according to the complexity of different areas, optimizing the computing resource usage while ensuring the accuracy of the model.

[0031] S2.2: Based on the equipment location information table and pipeline connection topology table in the virtual building skeleton model, calculate and construct the spatial position relationship matrix.

[0032] The construction process of the spatial position relationship matrix is ​​as follows: extract the device node coordinates and floor number from the device location information table, calculate the weighted Euclidean distance d between device i and device j ij , where vertical weighting coefficients are set for different distances between floors; pipeline type information is extracted from the pipeline connection topology table, and the connection coefficient c between device i and device j is determined based on the weight coefficients of different pipeline types ij ; Use the correlation strength calculation formula to construct an N×N spatial position relationship matrix M, where N is the total number of devices and the matrix element m ij Represents the spatial correlation strength between device i and device j.

[0033] Specifically, the calculation formula for association strength is as follows:

[0034] in, and are the weight coefficients of the distance term and the connection term respectively (0.6 and 0.4 in this embodiment), d0 is the distance attenuation coefficient (the value in this embodiment is 10 meters, which can be adjusted according to the actual building scale), and d ij is the weighted Euclidean distance between device i and device j, c ij is the weighted connection coefficient between device i and device j.

[0035] For example, the weight coefficient of pipeline type is set based on the importance of equipment energy flow and information flow, such as water supply pipeline weight 1.0, drainage pipeline weight 1.0, HVAC pipeline weight 1.5, power pipeline weight 1.2, network pipeline weight 0.8. In addition, when device i and device j are not connected, c ij When it is 0, and device i and device j are connected, the weighted connection coefficient c ij Calculated as follows:

[0036] Among them, K ij is the set of all pipeline types connecting device i and device j, is the weight coefficient corresponding to pipeline type k. When there are multiple types of pipeline connections between two devices, the maximum value of the weight coefficients of each pipeline type is taken as the weighted connection coefficient. This calculation method reflects the maximum impact of different types of pipelines on the relevance of equipment.

[0037] In specific implementation, the distance between devices is calculated using the three-dimensional Euclidean distance formula, and the vertical weighting coefficient is set to 1.5 to enhance the influence between floors. To optimize storage space, the device connection coefficient is stored in the adjacency matrix and represented by a sparse matrix. The association strength matrix uses its symmetry to store only the upper triangular part. All matrix elements are normalized and the value range is limited to the interval [0,1].

[0038] S2.3: Construct a spatial attention network based on the spatial position relationship matrix to generate spatial correlation weights between devices.

[0039] Specifically, the spatial position relationship matrix M is input into K attention heads, each of which contains a query matrix, a key matrix, and a value matrix; the attention score is calculated for each attention head; the outputs of the K attention heads are concatenated, and the spatial attention features are obtained through linear transformation; the spatial attention features are normalized to obtain the spatial correlation weight matrix between devices.

[0040] S2.4: Organize the standard data stream in time sequence and extract the dynamic time sequence characteristics of device operation through the temporal attention network.

[0041] Specifically, the standard data stream is grouped according to the device ID, and the time series data of each device is segmented using a sliding window with a window size of T; a bidirectional long short-term memory network is constructed for the segmented sequence to extract local time series patterns; the self-attention mechanism is used to calculate the importance weights of different time points, and the dynamic time series feature vector is obtained by weighted aggregation.

[0042] In the specific implementation, the sliding window size is set to 24 to capture the daily cycle pattern, and the step size is 1 to ensure fine-grained feature extraction. The BiLSTM network adopts a 2-layer structure, the hidden layer dimension is 128, and the dropout rate is set to 0.3 to prevent overfitting. The temporal attention mechanism introduces 32-dimensional relative position encoding to enhance the temporal modeling capability. The final feature vector dimension is 256, which contains short-term and long-term temporal information. The batch normalization layer is used to process features to accelerate training convergence.

[0043] S2.5: Fuse the spatial correlation weights and dynamic temporal features to generate a spatiotemporal fusion feature vector.

[0044] Specifically, the spatial correlation weight matrix W is multiplied by the dynamic timing feature vector of each device to obtain the timing feature considering the spatial correlation; the original timing feature and the spatial weighted feature are adaptively fused through the gating mechanism to output the spatiotemporal fusion feature vector.

[0045] S2.6: Organize the spatiotemporal fused feature vectors into a device feature matrix.

[0046] Furthermore, the spatiotemporal fusion feature vectors of each device are arranged in the order of device ID to construct an N×d-dimensional device feature matrix, where N is the number of devices and d is the feature dimension. Each row represents the spatiotemporal fusion feature of a device.

[0047] It should be noted that the present invention realizes the adaptive fusion of spatial relationships and temporal features between devices through a multi-head attention mechanism and a gated fusion mechanism, thereby improving the feature expression capability.

[0048] Preferably, the present invention realizes dynamic feature extraction of device operation status by introducing spatiotemporal attention mechanism and adaptive feature fusion strategy, overcoming the limitation of traditional methods that ignore spatial correlation between devices.

[0049] S3: Input the device feature matrix into the deep learning network, train and generate the energy consumption prediction model, and output the energy consumption prediction sequence.

[0050] Specifically, the following steps are included: S3.1: Slidingly split the device feature matrix according to a preset time window to generate a sample set containing historical operation data fragments, and divide the sample set into a training sample set and a verification sample set according to a preset ratio.

[0051] Each sample contains a sequence of equipment operation parameters and the corresponding actual energy consumption value, and the ratio of the training sample set to the verification sample set is set to 8:2.

[0052] S3.2: Construct a bidirectional long short-term memory network as the backbone network of the energy consumption prediction model, input the equipment operation parameter sequence of the training sample set into the bidirectional long short-term memory network, and extract the time series feature vector.

[0053] S3.3: A multi-head attention calculation unit is set in the output layer of the bidirectional long short-term memory network to calculate the weight of the time series feature vector and generate a fused feature vector.

[0054] S3.4: Input the fused feature vector into the fully connected layer to obtain the energy consumption prediction value, and calculate the mean square error loss value between the energy consumption prediction value and the actual energy consumption value.

[0055] S3.5: Based on the calculated mean square error loss value, the network parameters are updated using an adaptive learning rate optimization algorithm.

[0056] Specifically, the chain derivation method is used to calculate the gradient values ​​of the network parameters with respect to the loss function, and the back-propagation algorithm is used to update the gradient matrix layer by layer; the Adam optimizer parameters are initialized, and the first-order moment estimation coefficient, the second-order moment estimation coefficient, and the numerical stability coefficient are set; the exponentially weighted moving average of the gradient matrix is ​​calculated, and the first-order moment estimation vector is updated; the exponentially weighted moving average of the square of the gradient matrix is ​​calculated, and the second-order moment estimation vector is updated; the first-order moment estimation vector and the second-order moment estimation vector are bias-corrected to obtain the corrected first-order moment estimation vector and the second-order moment estimation vector; the adaptive learning rate is calculated according to the corrected moment estimation vector; the network parameters are updated based on the calculated adaptive learning rate and the corrected first-order moment estimation vector.

[0057] S3.6: Input the verification sample set into the bidirectional long short-term memory neural network, calculate the verification loss value, and use the early stopping mechanism based on the convergence of the verification loss to select the optimal model parameters to obtain the energy consumption prediction model.

[0058] Among them, the early stopping mechanism based on the convergence of verification loss is used to select the optimal model parameters, including: setting a verification loss monitoring window, recording the verification loss value after each round of training, calculating the moving average of the verification loss in the monitoring window, and constructing a smooth loss curve; calculating the verification loss change rate between adjacent rounds, judging the rising or falling trend of the loss based on the positive or negative change rate, and counting the number of consecutive rising rounds; judging whether the number of consecutive rising rounds exceeds the preset threshold, and triggering the early stopping condition when the threshold is exceeded, and recording the training round number when triggered; backtracking the historical verification loss records, locating the training round corresponding to the minimum verification loss value, and extracting the network parameters of this round as the optimal model parameters.

[0059] S3.6: Input the real-time device feature matrix into the energy consumption prediction model, and output the energy consumption prediction sequence for the next M time points.

[0060] Preferably, the energy consumption prediction accuracy is improved and the prediction lead time is extended by using a bidirectional long short-term memory network structure and a multi-head attention mechanism. The adaptive learning rate optimization algorithm is used to accelerate the model convergence speed.

[0061] S4: Construct a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generate a fault warning priority list based on the equipment weight coefficient.

[0062] Specifically, the fault warning priority list acquisition flow chart is as follows: Figure 3 As shown, the following steps are included: S4.1: Use the differential calculation unit to perform time series differential operation on the energy consumption prediction sequence to obtain an energy consumption change trend array.

[0063] Specifically, a time window matrix is ​​constructed, and the energy consumption prediction sequence is segmented according to a preset window length N, wherein each time window contains the predicted energy consumption values ​​of N adjacent time points; a first-order difference operation is performed on the time window matrix, and the difference between adjacent predicted energy consumption values ​​in each time window is calculated to generate an initial difference array; a median filter is performed on the initial difference array to eliminate the sudden change interference generated in the difference process and generate a smooth difference array; the smooth difference array is multiplied by the time step correction coefficient to obtain an energy consumption change rate array per unit time; a wavelet transform is used on the energy consumption change rate array to extract the main trend components of the energy consumption change, filter out high-frequency noise, and obtain an energy consumption change trend array.

[0064] S4.2: Call the time series pattern recognition module to extract features from the energy consumption change trend array to obtain the fluctuation characteristic value, trend slope value and maximum energy consumption difference.

[0065] Specifically, the energy consumption change trend array is resampled according to a fixed sampling frequency to generate equally spaced time series samples to ensure the time consistency of data analysis; a sliding window variance calculator is used to calculate the local variance value of the equally spaced time series samples, and the variance peak is selected as the fluctuation characteristic value; the least squares fitting module is called to perform linear regression operations on the equally spaced time series samples, and the slope of the fitting line is extracted as the trend slope value; based on the dynamic programming algorithm, the maximum difference between adjacent data points is searched in the equally spaced time series samples, and the difference is recorded as the maximum energy consumption difference; the fluctuation characteristic value, trend slope value and maximum energy consumption difference are combined into a feature vector for subsequent health risk score calculation.

[0066] S4.3: Input the fluctuation characteristic value, trend slope value and maximum energy consumption difference value into the pre-trained multi-layer perceptron and output the equipment health risk score.

[0067] Specifically, the fluctuation characteristic value, trend slope value and maximum energy consumption difference are normalized to generate a feature normalized vector, and the normalization adopts the maximum and minimum value standardization method; a three-layer perceptron network structure is constructed, the first layer contains 8 neurons, the second layer contains 4 neurons, and the output layer contains 1 neuron, and the layers are connected by ReLU activation function; the feature normalized vector is input to the first layer neurons, and the first layer output vector is calculated by weight matrix W1 and bias vector b1; the first layer output vector is processed by ReLU activation function and input to the second layer neurons, and the second layer output vector is calculated by weight matrix W2 and bias vector b2; the second layer output vector is processed by ReLU activation function and input to the output layer neurons, and the health risk score is calculated by weight matrix W3 and bias vector b3.

[0068] S4.4: Read the device weight coefficient in the device configuration database, multiply the device weight coefficient by the device health risk score, and obtain the device warning priority score.

[0069] S4.5: Execute a descending sorting algorithm to sort the equipment warning priority scores and generate an equipment fault warning priority list.

[0070] Among them, the equipment failure warning priority list includes equipment number, warning score, and warning level.

[0071] S5: Automatically generate maintenance work orders based on the fault warning priority list, assign work orders based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, push them to the operation and maintenance personnel, and feed back the work order execution results to the virtual building model for status update.

[0072] In a specific embodiment, the implementation process is as follows: S5.1: Call the maintenance knowledge base, match the corresponding standard maintenance process based on the equipment number and warning level, and generate an initial maintenance work order.

[0073] Specifically, read the equipment number, obtain the equipment type code, equipment model parameters, and equipment installation location information from the equipment basic database, and build an equipment feature identification set; use a feature vector matching algorithm to retrieve the maintenance knowledge base based on the equipment feature identification set, and extract the corresponding equipment maintenance procedure data, including standard maintenance steps, safety precautions, and maintenance tool requirements; input the warning level into the preset maintenance classification rule table, and extract the corresponding level of maintenance response requirements, including maintenance response time limit, maintenance personnel qualification requirements, and maintenance acceptance standards; extract the real-time operating parameters, historical maintenance records, and historical abnormal records of the corresponding equipment from the virtual building model, and generate a fault analysis report in combination with the maintenance procedure data, including abnormal parameter descriptions, possible fault causes, and recommended inspection items; match the spare parts management database based on the fault analysis report to generate a recommended spare parts list, including spare part number, spare part name, and recommended quantity; organize and integrate the maintenance procedure data, maintenance response requirements, fault analysis report, and recommended spare parts list, and format them according to the work order template format standard to generate an initial maintenance work order.

[0074] S5.2: Construct an equipment maintenance difficulty assessment model and calculate the equipment maintenance difficulty coefficient based on equipment type, fault complexity, and maintenance hours.

[0075] Firstly, a benchmark score table of equipment complexity is established based on the equipment type, and the benchmark scores of different types of equipment range from 0 to 100. Secondly, a fault complexity assessment matrix is ​​constructed, and the matrix dimensions include fault type, fault range, and fault level, and a weight coefficient is set for each dimension. Then, a random forest model is trained based on historical maintenance data to predict maintenance man-hours. Finally, the benchmark score, fault complexity score, and standardized maintenance man-hours are weighted and summed to obtain the equipment maintenance difficulty coefficient, which ranges from 0 to 1.

[0076] S5.3: Establish a skill evaluation system for operation and maintenance personnel to quantify their skill levels based on their work experience, professional qualifications, and historical maintenance success rates.

[0077] Firstly, a skill evaluation index system is constructed, including basic indicators such as years of work experience, professional qualification level, and certification status, as well as performance indicators such as historical maintenance completion rate and maintenance quality score. Secondly, the analytic hierarchy process is used to determine the weight of each indicator. Then, the comprehensive skill score is calculated based on the fuzzy comprehensive evaluation method. Finally, the Min-Max normalization method is used to map the comprehensive score to the range of 0-1 to obtain a standardized skill level score.

[0078] S5.4: Query the operation and maintenance personnel resource pool database, match the equipment maintenance difficulty coefficient with the skill level, use the task allocation algorithm to select operation and maintenance personnel who meet the qualification requirements, and establish a work order allocation table.

[0079] First, the system obtains the on-the-job status, workload and other information of the operation and maintenance personnel resource pool in real time; uses the cosine similarity algorithm to calculate the matching degree between the equipment maintenance difficulty coefficient and the personnel skill level; builds a task allocation priority queue based on the matching degree; combines the personnel workload and response time requirements, uses the Hungarian algorithm to perform optimal task allocation, and outputs the work order allocation table.

[0080] S5.5: Push the initial maintenance work order to the mobile terminal of the operation and maintenance personnel through the operation and maintenance management platform, trigger the work order to start the execution process, and record the time when the work order is received.

[0081] Among them, the operation and maintenance personnel’s mobile terminal displays work order details, equipment location navigation and maintenance operation guidance interface.

[0082] S5.6: Receive the work order execution data sent back by the operation and maintenance personnel through the mobile terminal, and update the status information of the corresponding device node in the virtual building model based on the work order execution data.

[0083] Among them, the work order execution data includes fault confirmation information, maintenance operation records and replacement parts list, and the status information includes equipment health status, maintenance timestamp and maintenance measure code.

[0084] S5.7: Monitor the operating status of the equipment after maintenance, collect the performance data of the first operating cycle after maintenance, compare and analyze it with the performance data before maintenance, and generate a maintenance effect evaluation report.

[0085] Among them, performance data includes equipment energy consumption indicators, operating parameters and status indicators.

[0086] S5.8: Based on the results of the maintenance effect evaluation report, determine whether it is necessary to start a secondary maintenance process. If secondary maintenance is required, return to S5.1 to regenerate the work order; otherwise, complete the current maintenance process.

[0087] Specifically, the system extracts the performance improvement and fault improvement indicators in the evaluation report, and compares the extracted indicators with the preset thresholds; if the indicators are lower than the thresholds, it is marked as requiring secondary maintenance and returns to S5.1 to regenerate the work order; if the indicators meet the standards, the equipment status is updated to normal, ending the maintenance process.

[0088] Furthermore, the present embodiment also provides a building operation and maintenance management system based on big data, including a data acquisition module, which is used to collect multi-source heterogeneous data through an Internet of Things sensor network, perform data standardization processing, and generate a standard data stream; a mapping module, which is used to build a virtual building model based on digital twin technology, map the standard data stream to the corresponding device nodes of the virtual building model, and use the spatiotemporal attention mechanism to extract the device operation characteristics to generate a device feature matrix; a prediction and analysis module, which is used to input the device feature matrix into a deep learning network, train and generate an energy consumption prediction model, and output an energy consumption prediction sequence; an evaluation and early warning module, which is used to build a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generate a fault warning priority list in combination with the equipment weight coefficient; a work order management module, which is used to automatically generate a maintenance work order according to the fault warning priority list, assign the work order based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, push it to the operation and maintenance personnel, and feed back the work order execution results to the virtual building model for status update.

[0089] In summary, the present invention realizes the real-time mapping of physical buildings and virtual models through the combination of IoT sensor networks and digital twin technology, provides an intuitive three-dimensional visualization interface, and greatly improves the efficiency of equipment management and fault location. The deep learning architecture of the bidirectional long short-term memory network combined with the multi-head attention mechanism can accurately capture the timing characteristics of equipment operation data, significantly improving the accuracy and prediction time span of energy consumption prediction. The fault warning mechanism based on equipment health score and weight coefficient, combined with the intelligent work order distribution system, realizes the timely discovery and rapid response of faults, and improves the initiative and pertinence of operation and maintenance work. Through the maintenance effect evaluation and secondary maintenance mechanism, a complete closed-loop management process is formed to ensure the quality of maintenance and continuously optimize the maintenance strategy, thereby improving the overall operation and maintenance management level.

[0090] Example 2, reference Figure 1 to Figure 3 , which is the second embodiment of the present invention, and this embodiment provides a building operation and maintenance management method based on big data. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0091] To verify the effectiveness of the building operation and maintenance management method, the research team selected a Grade A office building in Nanshan District, Shenzhen as the experimental object, with a construction area of ​​about 45,000 square meters and a total of 25 floors. The experimental period is from June 2023 to December 2023, covering two typical climate cycles in summer and winter.

[0092] In the early stages of the experiment, the research team deployed a comprehensive IoT sensor network, including 350 temperature sensors, 280 humidity sensors, 180 air quality sensors, 25 smart electricity meters, 25 smart water meters, and 160 crowd density sensors. The sensor sampling frequency was set to 5 minutes per time, and the edge computing device used Huawei Atlas 200 DK for data preprocessing and local caching. The data standardization process uses a unified UTC+8 time zone timestamp to ensure data synchronization.

[0093] In the process of constructing the virtual building model, the IFC format building information model exported by Autodesk Revit 2023 is used as the basic data. The 3D reconstruction accuracy reaches the centimeter level, and the model contains more than 12,000 geometric entities, covering all key equipment nodes. The meshing adopts tetrahedral mesh, the minimum mesh size is 0.1 meter, and the total number of meshes is about 2.8 million. The coordinate system adopts the WGS84 standard, and the positioning accuracy is better than 0.5 meter.

[0094] The training data set of the energy consumption prediction model contains 180 days of historical operation data, with a sampling interval of 5 minutes, totaling about 51,840 data samples. The bidirectional long short-term memory network adopts a two-layer structure, each layer contains 128 hidden units, and the multi-head attention mechanism sets 8 attention heads. The training uses the Adam optimizer, the initial learning rate is set to 0.001, the batch size is 64, and the training rounds are 200. The average prediction error on the validation set is 3.2%, and the prediction lead time can reach 48 hours.

[0095] As shown in the technical comparison table 1, the present invention is comprehensively compared with the existing building management system: Table 1 Performance comparison between the present invention and the existing building management system

[0096] From the comparison data, it can be seen that the method of the present invention is significantly better than the existing technical solutions in all key performance indicators. In particular, the prediction lead time and fault warning accuracy have been improved by 100% and 15.5% respectively, providing sufficient preparation time for preventive maintenance.

[0097] During the six-month experimental operation, the system processed 2,860 equipment abnormality warnings and generated 1,245 maintenance work orders. Through the hierarchical processing of the health scoring model, 85% of potential faults were handled in time before the impact expanded. The average repair time for equipment failures was reduced from 120 minutes to 45 minutes, and the one-time completion rate of repairs increased to 92%. The system also accurately tracked the energy consumption changes of various types of equipment. The monthly energy consumption of the central air-conditioning system was reduced by 18.5%, the energy consumption of the lighting system was reduced by 12.3%, and the energy consumption of the water supply and drainage system was reduced by 8.7%.

[0098] The experimental results show that this method not only significantly improves the efficiency of building operation and maintenance, but also achieves considerable energy-saving effects. Through the combination of digital twin technology and deep learning, the system can accurately predict equipment anomalies and ensure the optimal configuration of maintenance resources through an intelligent work order allocation mechanism. It is particularly noteworthy that the adaptive learning ability of the system enables the accuracy of the prediction model to continue to improve with the increase of running time. By the end of the experiment, the average error of energy consumption prediction has been reduced to less than 3.2%.

[0099] The follow-up evaluation of the maintenance effect shows that the performance recovery degree of the equipment after maintenance has reached an average of more than 95% of the factory standard, which is much higher than the 80% level of traditional maintenance methods. At the same time, the implementation of preventive maintenance has extended the average trouble-free operation time of the equipment by 25%, and it is expected to save more than 30% of the annual maintenance cost for the owner. These data fully prove the significant advantages and good economic benefits of this method in practical application.

[0100] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A building operation and maintenance management method based on big data, characterized by: include, Collect multi-source heterogeneous data through the IoT sensor network, perform data standardization processing, and generate standard data streams; A virtual building model is constructed based on digital twin technology, the standard data stream is mapped to the corresponding device nodes of the virtual building model, and the equipment operation characteristics are extracted by using a spatiotemporal attention mechanism to generate an equipment feature matrix; Inputting the device feature matrix into a deep learning network, training and generating an energy consumption prediction model, and outputting an energy consumption prediction sequence; Building a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generating a fault warning priority list in combination with the equipment weight coefficient; Automatically generate maintenance work orders according to the fault warning priority list, assign work orders based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, push them to the operation and maintenance personnel, and feed back the work order execution results to the virtual building model for status update; Generating a fault warning priority list comprises the following steps: The differential calculation unit is used to perform time series differential operation on the energy consumption prediction sequence to obtain the energy consumption change trend array; Calling a time series pattern recognition module to extract features from the energy consumption change trend array to obtain energy consumption change feature data; Inputting the energy consumption change characteristic data into a pre-trained multi-layer perceptron and outputting a device health risk score; Reading a device weight coefficient from a device configuration database, and multiplying the device weight coefficient by the device health risk score to obtain a device warning priority score; A descending sorting algorithm is executed to sort the equipment warning priority scores to generate an equipment fault warning priority list.

2. The building operation and maintenance management method based on big data according to claim 1, characterized in that: The construction of a virtual building model based on digital twin technology includes the following steps: Obtain the building information model data of the building, construct a three-dimensional geometric model of the building through a three-dimensional modeling engine, and generate a virtual building skeleton model; Based on the virtual building skeleton model, calculate and construct the spatial position relationship matrix; Building a spatial attention network based on the spatial position relationship matrix to generate spatial association weights between devices; Organize the standard data stream in time sequence and extract the dynamic time sequence characteristics of the device operation through the temporal attention network; Fusion of the spatial association weight and the dynamic time series feature to generate a spatiotemporal fusion feature vector; The spatiotemporal fusion feature vectors are organized into a device feature matrix.

3. The building operation and maintenance management method based on big data according to claim 1, characterized in that: The training and generation of the energy consumption prediction model comprises the following steps: Slidingly segmenting the device feature matrix according to a preset time window to generate a sample set, and dividing the sample set into a training sample set and a verification sample set according to a preset ratio; Constructing a bidirectional long short-term memory network as the backbone network of the energy consumption prediction model, inputting the training sample set into the bidirectional long short-term memory network, and extracting a time series feature vector; A multi-head attention calculation unit is set in the output layer of the bidirectional long short-term memory network to generate a fusion feature vector; Input the fused feature vector into the fully connected layer to obtain the energy consumption prediction value, and calculate the mean square error loss value between the energy consumption prediction value and the actual energy consumption value; According to the calculated mean square error loss value, the adaptive learning rate optimization algorithm is used to update the network parameters; The verification sample set is input into the bidirectional long short-term memory neural network, the verification loss value is calculated, and the optimal model parameters are selected by the early stopping mechanism based on the convergence of the verification loss to obtain the energy consumption prediction model; The real-time device feature matrix is ​​input into the energy consumption prediction model, and an energy consumption prediction sequence is output.

4. The building operation and maintenance management method based on big data according to claim 1, characterized in that: The energy consumption change characteristic data includes a fluctuation characteristic value, a trend slope value and a maximum energy consumption difference value.

5. The building operation and maintenance management method based on big data according to claim 1, characterized in that: Automatically generating a maintenance work order according to the fault warning priority list includes: Call the maintenance knowledge base, match the corresponding standard maintenance process based on the equipment number and warning level, and generate an initial maintenance work order; Construct an equipment maintenance difficulty assessment model to calculate the equipment maintenance difficulty coefficient based on equipment type, fault complexity, and maintenance hours; Establish a skill evaluation system for operation and maintenance personnel to quantify their skill levels based on their work experience, professional qualifications, and historical maintenance success rates; Query the operation and maintenance personnel resource pool database, match the equipment maintenance difficulty coefficient with the skill level, use the task allocation algorithm to select operation and maintenance personnel who meet the qualification requirements, and establish a work order allocation table; Pushing the initial maintenance work order to the mobile terminal of the operation and maintenance personnel through the operation and maintenance management platform, triggering the work order to start the execution process, and recording the time when the work order was received; Receiving the work order execution data sent back by the operation and maintenance personnel through the mobile terminal, and updating the status information of the corresponding device node in the virtual building model based on the work order execution data; Monitor the operating status of equipment after maintenance, collect performance data of the first operating cycle after maintenance, compare and analyze the performance data before maintenance, and generate a maintenance effect evaluation report; Based on the results of the maintenance effect evaluation report, determine whether it is necessary to start a secondary maintenance process.

6. The building operation and maintenance management method based on big data according to claim 4, characterized in that: The method of obtaining the energy consumption change trend array comprises the following steps: Construct a time window matrix and segment the energy consumption prediction sequence according to the preset window length N; Performing a first-order difference operation on the time window matrix, calculating the difference between adjacent predicted energy consumption values ​​in each time window, and generating an initial difference array; Performing median filtering on the initial difference array to generate a smoothed difference array; Multiplying the smoothed difference array by the time step correction coefficient to obtain an array of energy consumption change rates per unit time; Wavelet transform is applied to the energy consumption change rate array to extract the main trend components of the energy consumption change and obtain the energy consumption change trend array.

7. The building operation and maintenance management method based on big data according to claim 2, characterized in that: The construction process of the virtual building skeleton model includes: Importing building information model data, wherein the building information model data includes a floor plan, an equipment layout diagram, and a pipeline distribution diagram; Calling a geometry conversion engine to parse the building information model data and output a standard geometry description language file; Parsing the standard geometric description language file, extracting building component features, and establishing a component feature table; Based on the standard geometric description language file, a mesh generation operation is performed to generate a three-dimensional mesh data structure; The three-dimensional grid data structure and the building component feature table are loaded, and a hierarchical reconstruction algorithm is called to assemble space components to generate a virtual building skeleton model.

8. A building operation and maintenance management system based on big data, based on the building operation and maintenance management method based on big data according to any one of claims 1 to 7, characterized in that: Also includes, The data acquisition module is used to collect multi-source heterogeneous data through the IoT sensor network, perform data standardization processing, and generate standard data streams; A mapping module, used to build a virtual building model based on digital twin technology, map the standard data stream to the corresponding device nodes of the virtual building model, extract device operation features using a spatiotemporal attention mechanism, and generate a device feature matrix; A prediction analysis module, used to input the device feature matrix into a deep learning network, train and generate an energy consumption prediction model, and output an energy consumption prediction sequence; An evaluation and warning module is used to construct a multi-layer perceptron health scoring model based on the energy consumption prediction sequence, and generate a fault warning priority list in combination with the equipment weight coefficient; The work order management module is used to automatically generate maintenance work orders according to the fault warning priority list, allocate work orders based on the equipment maintenance difficulty coefficient and the skill level of the operation and maintenance personnel, push them to the operation and maintenance personnel, and feed back the work order execution results to the virtual building model for status update.

Images