Progress management system and method for electric power engineering

Abstract

The invention discloses a progress management system and method for electric power engineering, and belongs to the field of electric power construction engineering, and the system comprises a data collection module which is used for obtaining real-time sensor data, construction log data and external data of a construction site; the feature engineering module is used for integrating the multi-source data of the data acquisition module based on a spatial-temporal feature fusion engine and constructing an enhanced feature matrix; the model processing module is used for outputting a progress deviation and an equipment fault probability based on the PhaseAware LSTM network; key features are selected according to the construction stage, structured prediction is carried out based on a random forest, and a progress deviation rate prediction value and a risk factor evaluation result are output; and the dynamic weight fusion module is used for dynamically distributing model weights based on the prediction results of the random forest and the LSTM, the real-time equipment state and the environment parameters, and generating a fused progress prediction value. According to the invention, high-precision progress prediction and risk early warning can be output.

Description

Technical Field

[0001] The present invention relates to the field of electric power construction engineering, and in particular to a progress management system and method for electric power engineering. Background Art

[0002] Power engineering refers to a system engineering project that builds power generation, transmission, transformation, distribution and other facilities through planning, design, construction and commissioning. Power engineering construction includes multiple stages such as civil construction, equipment installation and commissioning, and system integration and acceptance. It is easily affected by external factors such as weather, geology, and policies. Different construction stages are affected to different degrees by different factors that affect the construction progress. The existing construction progress management method only tracks the progress by visualizing the task sequence and duration on the timeline, and cannot promptly detect risks that affect the construction progress. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a progress management system and method for power engineering, which can dynamically adjust the weights to adapt to different construction stages and external conditions by integrating the progress prediction model of LSTM and random forest, and output high-precision progress prediction and risk warning.

[0004] On the one hand, an embodiment of the present invention provides a progress management system for power engineering, including: a data acquisition module, which is used to obtain real-time sensor data, construction log data and external data of a construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel positioning data; the construction log data includes construction stage, construction plan and working time record; the external data includes weather data and holiday data; a feature engineering module, which is used to integrate multi-source data of the data acquisition module based on a spatiotemporal feature fusion engine, extract spatiotemporal features, and construct an enhanced feature matrix according to equipment status features, spatial collaboration features, time dimension features and environmental correction features; a model processing module, which is used to output progress deviation and equipment failure probability in a future period of time based on a PhaseAwareLSTM network, combined with construction stages and weather features; select key features according to the construction stage, perform structured prediction based on random forest, and output a progress deviation rate prediction value and a risk factor evaluation result of the random forest; a dynamic weight fusion module, which is used to dynamically allocate model weights based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, and generate a fused progress prediction value.

[0005] According to some embodiments of the present invention, the data acquisition module includes: an equipment vibration data unit, which is used to collect equipment vibration waveforms through an acceleration sensor, eliminate pulse interference based on median filtering, and extract 0-500Hz frequency band energy distribution through fast Fourier transform to obtain a vibration spectrum feature set; a temperature data unit, which is used to collect equipment surface temperature through a temperature sensor on the surface of key equipment components, calculate the minute-level temperature change gradient based on sliding window statistics, and obtain a temperature abnormality state mark; a spatial position unit, which is used to track engineering vehicles and mobile equipment through GPS, identify abnormal residence points based on the DBSCAN clustering algorithm for GPS trajectory data, and obtain an equipment space heat map; map the UWB coordinates of construction personnel to a grid of a preset size through geographic hash coding to obtain a personnel space heat map; a construction log parsing unit, which is used to obtain a temperature map from B The IM system imports the process time nodes, obtains the personnel arrival records from the attendance system, and reads the stage acceptance results from the quality management system; parses the unstructured log text and extracts the discrete construction stage labels; analyzes the difference between the planned working hours and the actual working hours through timestamp alignment to obtain the working hour difference coefficient; constructs the process progress timeline to obtain the process progress time series; converts the discrete construction stage labels into stage feature vectors based on One-Hot encoding; the external data unit is used to obtain weather data and holiday arrangement information, and the weather data includes: precipitation, wind speed and visibility data; constructs a dynamic influencing factor calculation model to obtain the environmental correction parameter set; constructs a weather-process association matrix to obtain a weather influencing factor matrix; uses date type tags to distinguish between working days / holidays, generates construction compliance verification marks, and obtains holiday status marks.

[0006] According to some embodiments of the present invention, the feature engineering module includes: a spatiotemporal alignment unit, which is used to map the vibration spectrum, temperature abnormality status mark, personnel / equipment spatial heat map, stage feature vector, working time difference coefficient, process progress time series, weather influence factor matrix, holiday status mark and environmental correction parameters to a unified time axis; and map the personnel spatial heat map to a unified grid coordinate system.

[0007] According to some embodiments of the present invention, the feature engineering module includes: an equipment status feature unit, which is used to obtain the vibration spectrum, calculate the RMS mean and the maximum kurtosis within the unit time window, and extract the 1-100Hz low-frequency band energy ratio; obtain the temperature abnormality state mark, count the number of consecutive over-temperature periods, record it as the temperature abnormality duration, and calculate the temperature rise rate; obtain the equipment health index according to the RMS mean, the low-frequency band energy ratio and the temperature abnormality duration; a spatial collaboration feature unit, which is used to obtain the personnel space manpower map and calculate the standard deviation of the personnel density of each grid; obtain the equipment space thermal map and the process progress time, and calculate the equipment transfer distance per unit time; and calculate the transportation efficiency according to the actual distance to the destination, the planned distance and the weather impact coefficient in the weather impact factor matrix; time The dimensional feature unit is used to obtain the progress acceleration within the sliding window of the process progress time series calculation and mark the progress lag status; obtain the working time difference coefficient and the weather influence factor matrix, correct the effective working time, and generate the resource utilization rate based on the effective working time; the environmental correction feature unit is used to obtain the weather influence factor matrix, construct a composite weather influence factor, and when the composite weather influence factor is greater than the threshold, trigger the construction efficiency degradation mark; set the holiday correction coefficient to correct the efficiency index when construction is carried out on holidays; the enhanced feature matrix unit is used to construct an enhanced feature matrix based on RMS mean, low-frequency energy proportion, temperature rise rate, standard deviation of personnel density, transportation efficiency, progress acceleration, lag mark, effective working time, resource utilization, composite weather influence factor and holiday correction coefficient.

[0008] According to some embodiments of the present invention, the model processing module includes: a PhaseAwareLSTM prediction unit, including a PhaseAwareLSTM network, including: a phase-gated LSTM layer, a time attention layer, a feature attention layer and a multi-task output head; the phase-gated LSTM layer introduces a construction phase feature vector in the LSTM hidden state update and dynamically adjusts the memory unit weights; the time attention layer is used to calculate the importance weights of historical time steps; the feature attention layer is used to strengthen the weights of equipment-related features; the multi-task output head is used to output progress deviations and equipment failure probabilities in a future period of time.

[0009] According to some embodiments of the present invention, the model processing module includes: a random forest unit, which is used to obtain an enhanced feature matrix, dynamically adjust the feature subset according to the construction stage label through stage-aware feature screening and rolling importance evaluation, use a multi-granularity tree structure to parallel process the feature subset, and output the result through a collaboratively trained dual-task forest; the multi-granularity tree structure includes a main tree group and an expert subtree group, the main tree group is used to process global features, and the expert subtree group is a decision path for processing specific scenarios; the dual-task forest includes a regression forest and a classification forest; the regression forest is used to predict the progress deviation rate based on adaptive binning and information gain splitting rules, and generate a final prediction value according to a dynamic calibration formula; the classification forest is used to output multi-level risk tags through probability threshold division and composite judgment rules, and trace the contribution of key features.

[0010] According to some embodiments of the present invention, the dynamic weight fusion module is used to receive the progress deviation rate prediction value of the random forest, the timing prediction value of the PhaseAwareLSTM and the real-time operating condition parameters, dynamically adjust the weight distribution through the error feedback mechanism and feature correlation analysis, and finally output the fused project progress prediction value and the calibrated risk level; the real-time operating condition parameters include equipment status characteristics and environmental correction characteristics.

[0011] According to some embodiments of the present invention, the system also includes: a feedback calibration module, which is used to receive the prediction results of the dynamic weight fusion module and real-time engineering verification data, dynamically correct the model output through an online learning mechanism and error propagation analysis, and finally output a calibrated progress prediction value and an optimized risk probability distribution; the real-time engineering verification data includes actual progress deviations and risk event markers.

[0012] According to some embodiments of the present invention, the system further comprises: a visualization module for displaying project progress forecast, risk distribution and model decision basis in real time through a multi-dimensional graphics rendering engine and an interactive analysis interface.

[0013] A progress management system for power engineering according to an embodiment of the present invention includes at least the following beneficial effects: the embodiment of the present invention integrates the progress prediction model of LSTM and random forest, dynamically adjusts the weights to adapt to different construction stages and external conditions, and outputs high-precision progress predictions and risk warnings; through error-driven weight allocation and feature-related working condition adaptation, dynamic optimization of the prediction model is achieved to ensure the timeliness and reliability of risk warnings.

[0014] On the other hand, an embodiment of the present invention provides a progress management method for power engineering, comprising the following steps: S100, acquiring real-time sensor data, construction log data and external data of the construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel positioning data; the construction log data includes construction stage, construction plan and working time record; the external data includes weather data and holiday data; S200, integrating multi-source data based on the spatiotemporal feature fusion engine, extracting spatiotemporal features, and constructing an enhanced feature matrix based on equipment status features, spatial collaboration features, time dimension features and environmental correction features; S300, based on the PhaseAwareLSTM network, combining the construction stage and weather features, outputting the progress deviation and equipment failure probability in the future period; selecting key features according to the construction stage, performing structured prediction based on random forest, and outputting the progress deviation rate prediction value and risk factor evaluation result of random forest; S400, dynamically allocating model weights based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, and generating a fused progress prediction value.

[0015] A progress management method for power engineering according to an embodiment of the present invention includes at least the following beneficial effects: the method according to an embodiment of the present invention is used to implement any of the above-mentioned systems, so the method according to an embodiment of the present invention has all the beneficial effects possessed by the above-mentioned systems.

[0016] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0018] Figure 1 A schematic block diagram of modules of a system according to an embodiment of the present invention;

[0019] Figure 2 The figure is a flowchart of a method according to an embodiment of the present invention.

[0020] Reference numerals:

[0021] Data collection module 100, feature engineering module 200, model processing module 300, dynamic weight fusion module 400. DETAILED DESCRIPTION

[0022] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0023] In the description of the present invention, "several" means one or more, "more" means two or more, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0024] Glossary:

[0025] The PhaseAware LSTM network is an LSTM variant network optimized for multi-stage time series prediction scenarios. The present invention embeds the physical semantic features of the construction phase (Phase) to achieve dynamic adaptation of feature weights of different engineering phases and improve the working condition perception capability of progress prediction.

[0026] ‌Random Forest‌ is an ensemble learning algorithm based on decision trees. It significantly improves the robustness and generalization ability of the model by building multiple decision trees and aggregating their prediction results. Its core design combines the ‌Bootstrap aggregation (Bagging)‌ and ‌Random feature subspace selection‌ mechanisms, and is particularly suitable for classification and regression tasks of high-dimensional, heterogeneous data, and has shown excellent performance in scenarios such as project progress prediction and risk assessment.

[0027] Reference Figure 1 The embodiment of the present invention provides a progress management system for electric power engineering, comprising:

[0028] The data acquisition module 100 is used to obtain real-time sensor data, construction log data and external data at the construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel positioning data; the construction log data includes construction stages, construction plans and working hours records; the external data includes weather data and holiday data.

[0029] The feature engineering module 200 is used to integrate the multi-source data of the data acquisition module based on the spatiotemporal feature fusion engine, extract spatiotemporal features, and construct an enhanced feature matrix based on device status features, spatial collaboration features, time dimension features and environmental correction features.

[0030] The model processing module 300 is used to output the progress deviation and equipment failure probability in the future period based on the PhaseAwareLSTM network in combination with the construction stage and weather characteristics; select key features according to the construction stage, perform structured prediction based on random forest, and output the progress deviation rate prediction value and risk factor evaluation result of the random forest.

[0031] The dynamic weight fusion module 400 is used to dynamically allocate model weights based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, and generate a fused progress prediction value.

[0032] In some embodiments, the data acquisition module 100 of the embodiment of the present invention includes:

[0033] The equipment vibration data unit is used to collect the equipment vibration waveform through the acceleration sensor, eliminate pulse interference based on the median filter, and extract the energy distribution of the 0-500Hz frequency band through the fast Fourier transform to obtain the vibration spectrum feature set.

[0034] The temperature data unit is used to collect the surface temperature of the equipment through the temperature sensor on the surface of the key equipment components, calculate the minute-level temperature change gradient based on the sliding window statistics, and obtain the temperature abnormality status mark.

[0035] The spatial location unit is used to track engineering vehicles and mobile equipment through GPS, identify abnormal residence points based on the DBSCAN clustering algorithm for GPS trajectory data, and obtain the equipment space heat map; the UWB coordinates of the construction personnel are mapped to a grid of preset size through geographic hash coding to obtain the personnel space heat map.

[0036] The construction log parsing unit is used to import process time nodes from the BIM system, obtain personnel arrival records from the attendance system, and read stage acceptance results from the quality management system; parse unstructured log text and extract discrete construction stage labels; perform difference analysis between planned working hours and actual working hours through timestamp alignment to obtain working hour difference coefficients; construct process progress timelines to obtain process progress time series; and convert discrete construction stage labels into stage feature vectors based on One-Hot encoding.

[0037] The external data unit is used to obtain weather data and holiday schedule information. The weather data includes: precipitation, wind speed and visibility data; construct a dynamic influencing factor calculation model to obtain an environmental correction parameter set; construct a weather-process association matrix to obtain a weather influencing factor matrix; use date type tags to distinguish between working days / holidays, generate construction compliance verification marks, and obtain holiday status marks.

[0038] In some embodiments, the feature engineering module 200 of the embodiment of the present invention includes: a spatiotemporal alignment unit, which is used to map the vibration spectrum, temperature abnormality status mark, personnel / equipment spatial heat map, stage feature vector, working time difference coefficient, process progress time series, weather influence factor matrix, holiday status mark and environmental correction parameter to a unified time axis; and map the personnel spatial heat map to a unified grid coordinate system.

[0039] In some embodiments, the feature engineering module 200 of the embodiment of the present invention includes:

[0040] The equipment status feature unit is used to obtain the vibration spectrum, calculate the RMS mean and maximum kurtosis within the unit time window, and extract the energy proportion of the low-frequency band of 1-100Hz; obtain the temperature abnormality status mark, count the number of consecutive over-temperature periods, record it as the temperature abnormality duration, and calculate the temperature rise rate; obtain the equipment health index based on the RMS mean, low-frequency band energy proportion and temperature abnormality duration.

[0041] The spatial collaboration feature unit is used to obtain the spatial manpower map of personnel and calculate the standard deviation of the personnel density of each grid; obtain the spatial heat map of equipment and the process progress time, calculate the equipment transfer distance per unit time; and calculate the transportation efficiency based on the actual distance to the destination, the planned distance and the weather impact coefficient in the weather impact factor matrix.

[0042] The time dimension feature unit is used to obtain the process progress time series to calculate the progress acceleration within the sliding window and mark the progress lag status; obtain the working time difference coefficient and weather influence factor matrix, correct the effective working time, and generate resource utilization based on the effective working time.

[0043] The environmental correction feature unit is used to obtain the weather impact factor matrix and construct a composite weather impact factor. When the composite weather impact factor is greater than the threshold, the construction efficiency degradation mark is triggered; the holiday correction coefficient is set to correct the efficiency index when construction is carried out on holidays.

[0044] The enhanced feature matrix unit is used to construct an enhanced feature matrix based on RMS mean, low-frequency energy proportion, temperature rise rate, standard deviation of personnel density, transportation efficiency, progress acceleration, lag mark, effective working hours, resource utilization, composite weather influence factor and holiday correction coefficient.

[0045] In some embodiments, the model processing module 300 of the embodiment of the present invention includes: a PhaseAwareLSTM prediction unit, including a PhaseAwareLSTM network, including: a phase-gated LSTM layer, a time attention layer, a feature attention layer and a multi-task output head; the phase-gated LSTM layer introduces the construction phase feature vector in the LSTM hidden state update and dynamically adjusts the memory unit weight; the time attention layer is used to calculate the importance weight of the historical time step and focus on the key time period (such as 2 hours before the equipment abnormality); the feature attention layer is used to strengthen the weight of equipment-related features (vibration, temperature) and suppress low-correlation features (such as holiday marks); the multi-task output head includes a regression branch and a classification branch, which are used to output the progress deviation and equipment failure probability in the future period respectively.

[0046] The input data of the PhaseAwareLSTM network in this embodiment include: time series features: sensor data (vibration, temperature, etc.), environmental parameters (rainfall, wind speed); phase coding: construction phase label (earthwork / structure / decoration, One-Hot coding), which is mapped into a dense vector through the embedding layer. The designed gate enhancement unit of the PhaseAwareLSTM network injects the phase embedding vector into the input gate and forget gate of the traditional LSTM to obtain the phase-aware gating.

[0047] In some embodiments, the model processing module 300 of the embodiment of the present invention includes: a random forest unit, which is used to obtain an enhanced feature matrix, dynamically adjust the feature subset according to the construction stage label through stage-aware feature screening and rolling importance evaluation, and use a multi-granularity tree structure to parallel process the feature subset, and output the results through a collaboratively trained dual-task forest; the multi-granularity tree structure includes a main tree group and an expert subtree group, the main tree group is used to process global features, and the expert subtree group is a decision path for processing specific scenarios; the dual-task forest includes a regression forest and a classification forest; the regression forest is used to predict the progress deviation rate based on adaptive binning and information gain splitting rules, and generate the final prediction value according to the dynamic calibration formula; the classification forest is used to output multi-level risk tags through probability threshold division and composite judgment rules, and trace the contribution of key features.

[0048] The stage-aware feature screening in this embodiment is used to dynamically adjust the feature subset according to the construction stage label (One-Hot encoding), and the rolling importance evaluation is used to recalculate the feature importance (based on Gini impurity) at fixed time intervals, and eliminate features whose contribution is less than the threshold. The node splitting rules of this embodiment include: for numerical features, processing based on adaptive binning (selecting the best split point based on the KS test); for categorical features, multi-way splitting based on information gain ratio progress. The composite judgment rules of this embodiment include multiple preset rules to adjust the risk level. For example, if the equipment overheating mark = 1 and the risk probability ≥ 0.4, it is forced to upgrade to high risk.

[0049] The random forest module of this embodiment achieves efficient processing of structured features through dynamic feature selection and multi-granularity tree group design. Its dual-task architecture can simultaneously output quantitative progress prediction and multi-level risk labeling, which is significantly better than the time series model in terms of real-time and interpretability. It forms a fast-slow synergistic system with LSTM: random forest is responsible for transient risk assessment and root cause location, and LSTM focuses on long-term trend prediction. The two improve the reliability of comprehensive decision-making through a probabilistic fusion mechanism. Unused time series fluctuation features (such as vibration spectrum details) are reserved for LSTM to avoid overlapping model functions.

[0050] In some embodiments, the dynamic weight fusion module 400 of the embodiment of the present invention is used to receive the progress deviation rate prediction value of the random forest, the time series prediction value of the PhaseAwareLSTM and the real-time working condition parameters, dynamically adjust the weight distribution through the error feedback mechanism and feature correlation analysis, and finally output the fused project progress prediction value and the calibrated risk level. The real-time working condition parameters include equipment status characteristics and environmental correction characteristics.

[0051] The error feedback mechanism of this embodiment obtains the historical error matrix (the prediction error of each model in the past 24 hours) and calculates the relative error of the model in the sliding window to construct a dynamic error matrix; the model weight is allocated according to the inverse of the error. The feature correlation analysis of this embodiment quantifies the feature contribution through the Pearson correlation coefficient (numerical type) or mutual information (category type) in the sliding window, and adjusts the weight through the preset working condition adaptive rule. For example, when the equipment is abnormal, the vibration feature correlation fluctuates abnormally (ΔMI>40%), and the random forest weight is forced to drop to 0, and only LSTM output is used. This embodiment uses a risk level calibration mechanism to calibrate the risk level, and sets a difference threshold trigger mechanism based on the conflict detection protocol. When the difference between the predicted values ​​of the two models is greater than the threshold, it is determined to be a prediction conflict, the risk level is increased, and the confidence is reduced. This embodiment realizes dynamic optimization of the prediction system through error-driven weight allocation, feature-related working condition adaptation, and conflict calibration risk control, reduces the frequency of manual intervention in complex construction scenarios, and ensures the timeliness and reliability of risk warning.

[0052] In some embodiments, the system of the embodiment of the present invention also includes: a feedback calibration module, which is used to receive the prediction results of the dynamic weight fusion module and real-time engineering verification data, dynamically correct the model output through online learning mechanism and error propagation analysis, and finally output the calibrated progress prediction value and the optimized risk probability distribution; the real-time engineering verification data includes actual progress deviation and risk event markers.

[0053] In some embodiments, the system of the embodiment of the present invention also includes: a visualization module for displaying project progress forecasts, risk distribution and model decision basis in real time through a multi-dimensional graphics rendering engine and an interactive analysis interface.

[0054] Reference Figure 2 The embodiment of the present invention provides a progress management method for power engineering, comprising the following steps:

[0055] S100, acquiring real-time sensor data, construction log data and external data of the construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel location data; the construction log data includes construction stage, construction plan and working time records; the external data includes weather data and holiday data.

[0056] S200, based on the spatiotemporal feature fusion engine, integrate multi-source data, extract spatiotemporal features, and construct an enhanced feature matrix based on device status features, spatial collaboration features, time dimension features, and environmental correction features.

[0057] S300, based on the PhaseAwareLSTM network, combined with the construction stage and weather characteristics, outputs the progress deviation and equipment failure probability in the future period; selects key features according to the construction stage, performs structured prediction based on random forest, and outputs the random forest progress deviation rate prediction value and risk factor assessment results.

[0058] S400, based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, dynamically assigns model weights and generates fused progress prediction values.

[0059] Although specific embodiments are described herein, those of ordinary skill in the art will recognize that many other modifications or alternative embodiments are also within the scope of the present disclosure. For example, any of the functions and / or processing capabilities described in conjunction with a particular device or component may be performed by any other device or component. In addition, although various exemplary implementations and architectures have been described according to embodiments of the present disclosure, those of ordinary skill in the art will recognize that many other modifications to the exemplary implementations and architectures described herein are also within the scope of the present disclosure.

[0060] It will be appreciated by those skilled in the art that all or some of the steps and systems in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0061] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions without violating the spirit of the present application. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A progress management system for electric power engineering, characterized in that: include: A data acquisition module is used to obtain real-time sensor data, construction log data and external data at the construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel location data; the construction log data includes construction stage, construction plan and working time records; the external data includes weather data and holiday data; The feature engineering module is used to integrate the multi-source data of the data acquisition module based on the spatiotemporal feature fusion engine, extract spatiotemporal features, and construct an enhanced feature matrix based on device status features, spatial collaboration features, time dimension features, and environmental correction features; The model processing module is used to introduce the construction phase feature vector into the LSTM hidden state update through the phase-gated LSTM layer of the PhaseAwareLSTM network, dynamically adjust the memory unit weights, and output the progress deviation and equipment failure probability in the future period based on the enhanced feature matrix input to the PhaseAwareLSTM network; obtain the enhanced feature matrix, adjust the feature subset according to the construction phase label, obtain the key features, perform structured prediction based on random forest, and output the progress deviation rate prediction value and risk factor evaluation result of the random forest; The dynamic weight fusion module is used to dynamically assign model weights based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, and generate fused progress prediction values.

2. The progress management system for electric power engineering according to claim 1, characterized in that: The data acquisition module comprises: The equipment vibration data unit is used to collect the equipment vibration waveform through the acceleration sensor, eliminate pulse interference based on the median filter, and extract the energy distribution of the 0-500Hz frequency band through the fast Fourier transform to obtain the vibration spectrum feature set; The temperature data unit is used to collect the surface temperature of the equipment through the temperature sensor on the surface of the key equipment components, calculate the minute-level temperature change gradient based on the sliding window statistics, and obtain the temperature abnormality status mark; The spatial location unit is used to track engineering vehicles and mobile equipment through GPS, identify abnormal residence points based on the DBSCAN clustering algorithm for GPS trajectory data, and obtain the equipment space heat map; the UWB coordinates of construction personnel are mapped to a grid of preset size through geographic hash coding to obtain the personnel space heat map; The construction log parsing unit is used to import process time nodes from the BIM system, obtain personnel arrival records from the attendance system, and read stage acceptance results from the quality management system; parse unstructured log texts and extract discrete construction stage labels; perform difference analysis between planned working hours and actual working hours through timestamp alignment to obtain working hour difference coefficients; construct process progress timelines to obtain process progress time series; and convert discrete construction stage labels into stage feature vectors based on One-Hot encoding; The external data unit is used to obtain weather data and holiday schedule information, wherein the weather data includes: precipitation, wind speed and visibility data; construct a dynamic influencing factor calculation model to obtain an environmental correction parameter set; construct a weather-process association matrix to obtain a weather influencing factor matrix; use date type tags to distinguish between working days / holidays, generate construction compliance verification marks, and obtain holiday status marks.

3. The progress management system for electric power engineering according to claim 2, characterized in that: The feature engineering module includes: The space-time alignment unit is used to map the vibration spectrum, temperature anomaly status mark, personnel / equipment space heat map, stage characteristic vector, working time difference coefficient, process progress time series, weather influence factor matrix, holiday status mark and environmental correction parameter to a unified time axis; and map the personnel space heat map to a unified grid coordinate system.

4. The progress management system for electric power engineering according to claim 2, characterized in that: The feature engineering module includes: The equipment status feature unit is used to obtain the vibration spectrum, calculate the RMS mean and maximum kurtosis within the unit time window, and extract the energy proportion of the low-frequency band of 1-100Hz; obtain the temperature abnormality status mark, count the number of consecutive over-temperature periods, record it as the temperature abnormality duration, and calculate the temperature rise rate; obtain the equipment health index based on the RMS mean, low-frequency band energy proportion and temperature abnormality duration; The spatial collaboration feature unit is used to obtain the spatial manpower map of personnel and calculate the standard deviation of the personnel density of each grid; obtain the spatial heat map of equipment and the process progress time, calculate the equipment transfer distance per unit time; and calculate the transportation efficiency based on the actual distance to the destination, the planned distance and the weather influence coefficient in the weather influence factor matrix; The time dimension feature unit is used to obtain the process progress time series to calculate the progress acceleration within the sliding window and mark the progress lag status; obtain the working time difference coefficient and weather influence factor matrix, correct the effective working time, and generate resource utilization based on the effective working time; The environmental correction feature unit is used to obtain the weather impact factor matrix and construct a composite weather impact factor. When the composite weather impact factor is greater than the threshold, the construction efficiency degradation mark is triggered; the holiday correction coefficient is set to correct the efficiency index when construction is carried out during holidays; The enhanced feature matrix unit is used to construct an enhanced feature matrix based on RMS mean, low-frequency energy proportion, temperature rise rate, standard deviation of personnel density, transportation efficiency, progress acceleration, lag mark, effective working hours, resource utilization, composite weather influence factor and holiday correction coefficient.

5. The progress management system for electric power engineering according to claim 1, characterized in that: The model processing module includes: The PhaseAwareLSTM prediction unit includes a PhaseAwareLSTM network, including: a phase-gated LSTM layer, a time attention layer, a feature attention layer and a multi-task output head; the time attention layer is used to calculate the importance weight of historical time steps; the feature attention layer is used to strengthen the weight of equipment-related features; the multi-task output head is used to output the progress deviation and equipment failure probability in a future period of time.

6. The progress management system for electric power engineering according to claim 1, characterized in that: The model processing module includes: The random forest unit is used to obtain an enhanced feature matrix, dynamically adjust the feature subset according to the construction stage label through stage-aware feature screening and rolling importance evaluation, use a multi-granularity tree structure to process the feature subset in parallel, and output the result through a collaboratively trained dual-task forest; the multi-granularity tree structure includes a main tree group and an expert subtree group, the main tree group is used to process global features, and the expert subtree group is a decision path for processing specific scenarios; the dual-task forest includes a regression forest and a classification forest; the regression forest is used to predict the progress deviation rate based on adaptive binning and information gain splitting rules, and generate the final prediction value according to the dynamic calibration formula; the classification forest is used to output multi-level risk tags through probability threshold division and compound judgment rules, and trace the contribution of key features.

7. The progress management system for electric power engineering according to claim 1, characterized in that: The dynamic weight fusion module is used to receive the progress deviation rate prediction value of the random forest, the time series prediction value of the PhaseAwareLSTM and the real-time operating condition parameters, dynamically adjust the weight distribution through the error feedback mechanism and feature correlation analysis, and finally output the fused project progress prediction value and the calibrated risk level; the real-time operating condition parameters include equipment status characteristics and environmental correction characteristics.

8. The progress management system for electric power engineering according to claim 1, characterized in that: The system further comprises: The feedback calibration module is used to receive the prediction results of the dynamic weight fusion module and the real-time engineering verification data, dynamically correct the model output through the online learning mechanism and error propagation analysis, and finally output the calibrated progress prediction value and the optimized risk probability distribution; the real-time engineering verification data includes the actual progress deviation and the risk event mark.

9. The progress management system for electric power engineering according to claim 1, characterized in that: The system further comprises: The visualization module is used to display the project progress forecast, risk distribution and model decision basis in real time through a multi-dimensional graphics rendering engine and an interactive analysis interface.

10. A progress management method for power engineering, used in the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S100, acquiring real-time sensor data, construction log data and external data at the construction site; the real-time sensor data includes equipment vibration data, temperature data, material transportation trajectory data and personnel location data; the construction log data includes construction stage, construction plan and working time records; the external data includes weather data and holiday data; S200, integrating multi-source data based on the spatiotemporal feature fusion engine, extracting spatiotemporal features, and constructing an enhanced feature matrix based on device status features, spatial collaboration features, time dimension features, and environmental correction features; S300, based on the PhaseAwareLSTM network, combines the construction phase and weather characteristics to output the progress deviation and equipment failure probability in the future period; Select key features according to the construction stage, perform structured prediction based on random forest, and output the random forest progress deviation rate prediction value and risk factor assessment results; S400, based on the prediction results of random forest and LSTM, real-time equipment status and environmental parameters, dynamically assigns model weights and generates fused progress prediction values.

Images