Modularized rapid deployment system for new energy facilities

Abstract

The invention provides a new energy facility modularized rapid deployment system, which comprises a deployment planning decision module, a digital twin module, an execution control module, a dynamic environment adaptation module and an efficiency evaluation module, and is used for realizing rapid, efficient and reliable deployment of new energy facilities in a complex environment. The system generates a deployment scheme through multi-objective optimization, performs simulation verification and execution control by using digital twinning to realize closed-loop automatic deployment, dynamically adapts to environmental changes, evaluates deployment efficiency in the whole process, and continuously optimizes a deployment strategy. According to the method, rapid deployment, efficient operation and dynamic optimization of the new energy facilities can be realized, the deployment efficiency, the resource utilization rate and the system reliability are improved, the deployment cost and risk are reduced, and the adaptability and stability of the new energy facilities in a complex environment are enhanced.

Description

Technical Field

[0001] This invention relates to the field of new energy facility deployment technology, and more specifically to a modular rapid deployment system for new energy facilities. Background Technology

[0002] With the rapid development of new energy sources, the demand for their deployment is increasing daily. Traditional deployment methods for new energy facilities typically rely on manual planning and on-site construction, which presents numerous problems. First, deployment planning lacks systematic approaches and optimization methods, making it difficult to comprehensively consider factors such as cost, time, and resource utilization, resulting in low deployment efficiency. Second, on-site construction is greatly affected by the environment, lacking an effective dynamic adjustment mechanism; once the environment changes, deployment progress and quality are difficult to guarantee. Furthermore, the deployment process lacks real-time monitoring and anomaly handling capabilities, making equipment installation and commissioning prone to malfunctions, which are difficult to quickly locate and resolve. Finally, after deployment, there is a lack of comprehensive performance evaluation and continuous optimization mechanisms, making it difficult to quantitatively analyze and improve the deployment effects. Summary of the Invention

[0003] This invention provides a modular rapid deployment system for new energy facilities, comprising: The deployment planning and decision-making module is used to receive geospatial data of the deployment environment, specification parameter data of new energy facilities, and timeliness requirements data of deployment tasks. It solves the optimal deployment scheme by establishing a multi-objective optimization model and considering various practical constraints. Deploy a digital twin module to build a virtual deployment environment based on the optimal deployment scheme, including terrain features, equipment physical characteristics and operating logic, and verify the feasibility and robustness of the deployment scheme under various working conditions through simulation; The deployment execution control module is used to decompose the verified deployment scheme into a sequence of control instructions with time sequence and logical relationship, and to adaptively adjust the control parameters by monitoring the deviation between the execution status and the expected status in real time. The dynamic environment adaptation module is used to continuously collect environmental parameter data and equipment operating status data at the deployment site, identify significant changes in environmental conditions through a change point detection algorithm, and dynamically update the deployment strategy based on an online learning mechanism. The deployment performance evaluation module is used to collect data from the beginning to the end of the deployment process, establish a multi-dimensional evaluation system including efficiency indicators, quality indicators, cost indicators and reliability indicators, and generate optimization suggestions through data analysis.

[0004] Furthermore, the deployment planning decision module executes a decision-making process including the following steps: Collect geospatial data on the latitude and longitude coordinates, altitude, terrain slope, and land cover type of the deployment area; obtain specification parameters of the size, weight, power output characteristics, and interface standards of the new energy facilities to be deployed; and receive deployment task requirements data specified by the user, such as deployment completion time, budget limit, and performance indicators. Establish a multi-objective optimization model that simultaneously considers minimizing deployment costs, minimizing deployment time, and maximizing resource utilization. Deployment costs include equipment procurement costs, transportation costs, installation costs, and maintenance costs. Deployment time includes equipment transportation time, on-site assembly time, system debugging time, and trial operation time. Resource utilization includes land utilization rate, equipment utilization rate, and energy conversion efficiency. An improved non-dominated sorting genetic algorithm is used to solve the multi-objective optimization model. This algorithm improves search efficiency and solution quality by introducing an elite retention strategy and an adaptive crossover mutation operator, and finally outputs a set of Pareto optimal deployment schemes. Based on a decision-making method combining entropy weighting and ideal point methods, the deployment scheme that best meets the actual needs is selected from the Pareto optimal solution set. The entropy weighting method is used to objectively determine the weight of each indicator, while the ideal point method is used to evaluate the degree of closeness of each scheme to the ideal scheme.

[0005] Furthermore, the solution process of the improved non-dominated sorting genetic algorithm specifically includes: The Latin hypercube sampling method is used to generate an initial population with good spatial distribution characteristics within the feasible region of the decision variables, where the decision variables include equipment layout coordinates, number of equipment, connection topology and installation sequence. Based on the Pareto dominance relationship, individuals in the population are ordered into non-dominated layers, and the population is divided into multiple non-dominated layers. The first layer consists of individuals that are not dominated by any other individuals, the second layer consists of individuals that are dominated only by individuals in the first layer, and so on. The crowding distance of individuals in the same non-dominated layer is calculated. This distance is used to assess the distribution density of individuals in the target space by measuring the sum of the distances between an individual and its neighboring individuals in each objective function direction. A binary tournament selection method is used to select parent individuals from the current population. The selection criteria prioritize individuals with high non-dominant ranking, and within the same ranking, individuals with large crowding distance are selected to maintain population diversity. Simulated binary crossover and polynomial mutation operations are performed on the selected parent individuals to generate offspring individuals, wherein the crossover probability and mutation probability are adaptively adjusted according to the population iteration generation and the individual distribution. The parent and offspring populations are merged, and a new generation of population is generated through environmental selection. Environmental selection takes into account non-dominant ranking and crowding distance to ensure that superior individuals are preserved while maintaining population diversity.

[0006] Furthermore, the deployed digital twin module performs a simulation verification process including the following steps: A topographic model is constructed based on a high-precision digital elevation model of the deployment area, geological survey data, and historical meteorological data; a physical model of the equipment is constructed based on a 3D CAD model of the new energy facilities, material properties, and dynamic parameters; and an operation logic model is constructed based on system control logic and operation strategies. The optimal deployment scheme is transformed into a resolvable simulation script for the digital twin environment. This script defines in detail the device installation location, installation sequence, connection relationship and control parameters, and is injected into the digital twin model through the application programming interface. Multiple simulation experiments are executed concurrently in a digital twin environment. Each experiment simulates different environmental conditions and operating scenarios, including normal operating conditions, extreme weather conditions, and equipment failures, to comprehensively evaluate the adaptability of the deployment plan. Time series data of key system performance indicators during the simulation process are collected, including energy output power, equipment operating status, system efficiency and stability indicators, and the data are preprocessed and feature extracted. Statistical hypothesis testing methods are used to analyze simulation results and determine whether the deployment scheme meets the preset performance requirements and safety standards, including mean test, variance test and distribution fit test.

[0007] Furthermore, the deployed digital twin module also performs a global sensitivity analysis including the following steps: Determine the input parameters in the digital twin model, including equipment installation location deviation, ambient temperature variation range, wind speed fluctuation characteristics, solar radiation intensity and load power variation pattern, and specify a reasonable value range and probability distribution for each parameter; The Sobol sequence sampling method generates a low-bias set of sampling points in the input parameter space. This method improves the accuracy of sensitivity analysis by ensuring the uniform distribution of sampling points in the parameter space. The digital twin model is run with each sampling point as input, and the corresponding system output response is recorded, including total power generation, system efficiency, equipment load rate and operation stability indicators. The first-order sensitivity index and the total sensitivity index of each input parameter are calculated based on the principle of variance decomposition. The first-order sensitivity index measures the independent contribution of a single parameter to the output response, while the total sensitivity index measures the total contribution of the parameter itself and its interaction with other parameters to the output response. All input parameters are sorted according to their sensitivity index to identify the key parameters that have the most significant impact on system performance, providing a focus for optimizing deployment plans and making on-site adjustments.

[0008] Furthermore, the deployment execution control module performs the following steps: The verified deployment scheme is parsed into a sequence of control instructions with strict time order and logical dependencies. Each control instruction clearly specifies the action to be performed, the target location, the action parameters, and the completion criteria. At the beginning of each control cycle, based on the currently collected system state data, a prediction model is used to estimate the evolution of the system state over a future period of time. The prediction model comprehensively considers the dynamic characteristics of the equipment, environmental disturbance factors, and actuator response delay. Solve the rolling time-domain optimization problem with multiple constraints. The goal is to minimize the deviation between the predicted output and the desired trajectory while minimizing the control energy consumption. The constraints include control magnitude constraints, control increment constraints, and output magnitude constraints. The first control quantity in the control command sequence obtained by optimization is sent to the actuator to control the field equipment to complete the specified actions, including equipment handling, position adjustment, connection and installation, and parameter configuration. The actual output response of the system is collected in real time by a sensor network. The deviation between the actual response and the predicted response is calculated. This deviation is used to correct the state estimation of the prediction model and enhance the robustness of the model in environments with uncertainty.

[0009] Furthermore, the deployment execution control module also performs the following steps: Multidimensional time-series data is collected by deploying visual, mechanical, and position sensors on-site. Spatial features are extracted from image data using convolutional neural networks, and temporal-dependent features are extracted from time-series data using long short-term memory networks. The extracted spatial and temporal features are fused and input into a fully connected neural network for anomaly identification. Anomaly types include equipment positioning deviation exceeding limits, installation connection failure, actuator jamming, and excessive environmental interference. When an anomaly is detected, a root cause analysis process based on graph neural networks is initiated. By constructing a device correlation graph and a fault propagation model, the root cause and scope of impact of the anomaly are located. Anomaly handling strategies are generated based on deep reinforcement learning algorithms. The agent selects the optimal handling action according to the current system state and anomaly type, including adjusting installation parameters, switching to backup equipment, activating security protection mechanisms, and notifying human intervention. Establish an exception handling knowledge base to record the event description, handling measures, and handling effects of each exception, and provide handling references for similar exceptions in the future through case reasoning methods.

[0010] Furthermore, the dynamic environment adaptation module performs the following steps: Time-series data of environmental parameters, including temperature, humidity, wind speed, light intensity, soil bearing capacity, and equipment vibration frequency, are continuously collected through weather stations, geological monitoring instruments, and equipment status sensors deployed on site. The cumulative sum algorithm is applied to process the time series of environmental parameters. By comparing the calculated decision function with the preset threshold, abrupt changes in environmental parameters are detected, and abnormal environmental changes are discovered in a timely manner. The Kalman filter algorithm is used to estimate and predict environmental parameters. By combining the system dynamics model and real-time observation data, the accuracy and reliability of environmental parameter prediction are improved. When a significant change in the environment is detected, a strategy adjustment mechanism is activated. Based on the current environmental conditions and system state, a fuzzy inference system is used to assess the impact of the environmental change on the deployment process. The Q-learning algorithm is applied to adjust deployment strategy parameters online. The agent obtains reward signals by interacting with the environment and continuously optimizes its decision-making strategy in changing environments to ensure the smooth progress of the deployment process.

[0011] Furthermore, the deployment performance evaluation module performs the following steps: Extract time-dimensional data from the deployment process records, including equipment transportation time, on-site assembly time, system debugging time, and trial operation time, and calculate deployment efficiency indicators such as deployment cycle completion rate, process connection efficiency, and resource scheduling efficiency; Extract resource dimension data from resource utilization records, including equipment usage, manpower input, and material consumption, and calculate resource utilization indicators such as equipment utilization rate, manpower load rate, and material loss rate. Extract quality dimension data from system operation records, including system power generation, energy conversion efficiency, operational stability and failure rate, and calculate deployment quality indicators such as performance compliance rate, system reliability and maintenance convenience; The weight coefficients of each dimension index are determined by a combined weighting method. The subjective weights are obtained through the analytic hierarchy process to reflect expert experience and judgment, while the objective weights are obtained through the entropy weighting method to reflect the information content of the data itself. The comprehensive performance score is calculated by multiplying the standardized values ​​of each dimension index by their corresponding weights and then summing the results to obtain a quantitative assessment of deployment performance. Based on the evaluation results, specific improvement suggestions are generated, and optimization measures are proposed for the less efficient links, including process reengineering, resource allocation optimization, and technical solution improvement.

[0012] Furthermore, the deployment performance evaluation module also performs a continuous improvement process including the following steps: During the planning phase, improvement opportunities are identified based on historical deployment effectiveness evaluation results. Causal analysis tools are used to identify key factors affecting effectiveness, and specific, measurable, achievable, relevant, and time-bound improvement goals are set. During the implementation phase, a detailed implementation plan for improvement measures should be developed, clearly defining the division of responsibilities, resource allocation, and timelines. The implementation status and intermediate results of the improvement measures should be fully recorded during the implementation process. During the inspection phase, data on the improved deployment process are collected, the changes in performance indicators before and after the improvement are compared, statistical testing methods are used to verify the significance of the improvement effect, and the reasons for not achieving the expected goals are analyzed. During the handling phase, proven effective improvement measures are incorporated into standard operating procedures and promoted throughout the organization through training and culture building. Meanwhile, unresolved issues are carried over to the next improvement cycle. Establish a quantitative evaluation model for improvement effects. This model comprehensively considers the degree of improvement in deployment efficiency, the degree of improvement in deployment quality, the degree of cost reduction, and the degree of reliability improvement, and calculates the comprehensive score of improvement effects by weighted summation.

[0013] The embodiments of the present invention have at least the following beneficial effects: 1. The introduction of a digital twin module enables the system to comprehensively simulate and verify deployment schemes in a virtual environment, allowing for early assessment of their feasibility and robustness under various operating conditions. This verification method, based on high-precision models and concurrent simulation, effectively reduces the risks and uncertainties of on-site deployment, avoids repeated construction and equipment damage caused by unreasonable schemes, and improves the overall reliability and stability of the system.

[0014] 2. The combination of the dynamic environment adaptation module and the deployment performance evaluation module enables real-time monitoring and dynamic adjustment of on-site environmental parameters and equipment operating status. Simultaneously, it conducts multi-dimensional performance evaluations throughout the deployment process and generates optimization suggestions. This mechanism allows the system to quickly respond to environmental changes, adjust deployment strategies promptly, ensure smooth deployment, and continuously optimize deployment results. It solves the problem of traditional deployment methods lacking dynamic adjustment and continuous improvement capabilities, thus enhancing the system's adaptability and deployment quality. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. The following drawings are not intentionally drawn to scale to actual size; their focus is on illustrating the main points of this disclosure.

[0016] Figure 1 This is a schematic diagram of the structure of a modular rapid deployment system for new energy facilities provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are also within the scope of protection of this disclosure.

[0018] The following is for reference. Figure 1 , Figure 1 This is a schematic diagram of a modular rapid deployment system for new energy facilities provided in an embodiment of the present invention. Figure 1 As shown, a modular rapid deployment system for new energy facilities includes: The deployment planning and decision-making module 101 is used to receive geospatial data of the deployment environment, specification parameter data of new energy facilities, and timeliness requirement data of deployment tasks. It solves the optimal deployment scheme by establishing a multi-objective optimization model and considering various practical constraints. Deploy digital twin module 102 to construct a virtual deployment environment based on the optimal deployment scheme, including terrain features, equipment physical characteristics and operating logic, and verify the feasibility and robustness of the deployment scheme under various working conditions through simulation; The deployment execution control module 103 is used to decompose the verified deployment scheme into a sequence of control instructions with time sequence and logical relationship, and to adaptively adjust the control parameters by monitoring the deviation between the execution status and the expected status in real time. The dynamic environment adaptation module 104 is used to continuously collect environmental parameter data and equipment operating status data at the deployment site, identify significant changes in environmental conditions through a change point detection algorithm, and dynamically update the deployment strategy based on an online learning mechanism. The deployment performance evaluation module 105 is used to collect data from the beginning to the end of the deployment process, establish a multi-dimensional evaluation system including efficiency indicators, quality indicators, cost indicators and reliability indicators, and generate optimization suggestions through data analysis.

[0019] Specifically, the geospatial data received by the deployment planning and decision-making module includes the latitude and longitude coordinates, altitude, terrain slope, and land cover type of the deployment area. This data is used to construct a geographic model of the deployment area. Specifications of the new energy facilities cover equipment dimensions, weight, power output characteristics, and interface standards. These parameters are used to determine the physical characteristics and compatibility of the equipment. Timeliness requirements for the deployment task include user-specified deployment completion time, budget limits, and performance indicators. These requirements constrain the economy and efficiency of the deployment plan.

[0020] In the multi-objective optimization model, deployment costs include equipment procurement costs, transportation costs, installation costs, and maintenance costs; deployment time includes equipment transportation time, on-site assembly time, system debugging time, and trial operation time; resource utilization includes land utilization rate, equipment utilization rate, and energy conversion efficiency. By setting these specific parameters, the system can comprehensively consider various factors in the deployment process, ensuring the optimality of the deployment plan.

[0021] The deployment planning and decision-making module employs an improved non-dominated sorting genetic algorithm to solve the multi-objective optimization model. Elite retention strategies and adaptive crossover and mutation operators are used to improve search efficiency and solution quality.

[0022] Furthermore, the deployment planning and decision-making module first collects geospatial data such as the latitude and longitude coordinates, altitude, terrain slope, and land cover type of the deployment area. This data is used to construct a geographic model of the deployment area to determine the equipment's installation location and environmental adaptability. Specification parameters include the equipment's dimensions, weight, power output characteristics, and interface standards to ensure physical compatibility and technical compatibility. Deployment task requirements involve user-specified deployment completion time, budget limits, and performance indicators to constrain the economy and efficiency of the deployment plan. In the multi-objective optimization model, deployment costs include equipment procurement costs, transportation costs, installation costs, and maintenance costs; deployment time includes equipment transportation time, on-site assembly time, system debugging time, and trial operation time; resource utilization includes land utilization rate, equipment utilization rate, and energy conversion efficiency. Through the setting of these specific parameters, the system can comprehensively consider various factors in the deployment process to ensure the optimality of the deployment plan.

[0023] The multi-objective optimization model is solved using an improved non-dominated sorting genetic algorithm. By introducing an elite retention strategy and an adaptive crossover and mutation operator, the search efficiency and solution quality are improved. Specifically, the algorithm first generates an initial population using the Latin hypercube sampling method. Individuals in the population represent different deployment schemes, and decision variables include equipment layout coordinates, number of equipment, connection topology, and installation order. During optimization, the population is non-dominatedly sorted based on Pareto dominance, and the crowding distance of individuals is calculated to maintain population diversity. Offspring individuals are generated by simulating binary crossover and polynomial mutation operations, with crossover and mutation probabilities adaptively adjusted according to the number of generations and individual distribution. Finally, an environment selection operation generates a new generation of population, ensuring that superior individuals are retained. Furthermore, the system combines entropy weighting and ideal point methods to select the deployment scheme that best meets the actual needs from the Pareto optimal solution set. The entropy weighting method is used to objectively determine the weights of each indicator, while the ideal point method is used to evaluate the closeness of each scheme to the ideal scheme.

[0024] Furthermore, the solution process of the improved non-dominated sorting genetic algorithm specifically includes: The Latin hypercube sampling method is used to generate an initial population with good spatial distribution characteristics within the feasible region of the decision variables, where the decision variables include equipment layout coordinates, number of equipment, connection topology and installation sequence. Based on the Pareto dominance relationship, individuals in the population are ordered into non-dominated layers, and the population is divided into multiple non-dominated layers. The first layer consists of individuals that are not dominated by any other individuals, the second layer consists of individuals that are dominated only by individuals in the first layer, and so on. The crowding distance of individuals in the same non-dominated layer is calculated. This distance is used to assess the distribution density of individuals in the target space by measuring the sum of the distances between an individual and its neighboring individuals in each objective function direction. A binary tournament selection method is used to select parent individuals from the current population. The selection criteria prioritize individuals with high non-dominant ranking, and within the same ranking, individuals with large crowding distance are selected to maintain population diversity. Simulated binary crossover and polynomial mutation operations are performed on the selected parent individuals to generate offspring individuals, wherein the crossover probability and mutation probability are adaptively adjusted according to the population iteration generation and the individual distribution. The parent and offspring populations are merged, and a new generation of population is generated through environmental selection. Environmental selection takes into account non-dominant ranking and crowding distance to ensure that superior individuals are preserved while maintaining population diversity.

[0025] Specifically, the solution process of the improved non-dominated sorting genetic algorithm includes: generating the initial population using the Latin hypercube sampling method. Latin hypercube sampling is an efficient sampling method that can generate sample points with good spatial distribution characteristics within the feasible region of the decision variables, thereby improving the quality of the initial population. Decision variables include equipment layout coordinates, number of devices, connection topology, and installation order, which directly determine the specific form of the deployment scheme.

[0026] Based on Pareto dominance, individuals in the population are non-dominated and ordered, dividing the population into multiple non-dominated layers. Non-dominated ordering is an important concept in multi-objective optimization, used to distinguish the relative merits of individuals in multi-objective optimization. Individuals in the same non-dominated layer have similar optimization levels; in this case, crowding distance is calculated to assess the distribution density of individuals in the objective space, thus maintaining population diversity. A binary tournament selection method is used to select parent individuals from the current population, and simulated binary crossover and polynomial mutation operations are performed to generate offspring individuals. The crossover and mutation probabilities are adaptively adjusted according to the number of generations and the distribution of individuals in the population to balance the capabilities of global and local search.

[0027] The improved non-dominated sorting genetic algorithm generates a new generation of population by comprehensively considering the non-dominated sorting level and crowding distance through an environmental selection operation. The purpose of environmental selection is to maintain population diversity while preserving superior individuals and avoiding the algorithm from getting trapped in local optima.

[0028] Specifically, the parent and offspring populations are merged, and then filtered based on non-dominated ranking and crowding distance to ensure that the new generation population includes both high-quality solutions and covers different regions of the target space. Furthermore, after generating the Pareto optimal solution set, the algorithm combines entropy weighting and ideal point methods to select the final solution. Entropy weighting is an objective weight determination method based on information entropy, used to assign weights according to the information content of each target indicator, thereby avoiding the influence of human factors on decision-making. Ideal point method selects the deployment scheme that best meets the actual needs by evaluating how close each scheme is to the ideal scheme. This method comprehensively considers the trade-offs between multiple objectives, ensuring that the final selected scheme performs well in multiple aspects.

[0029] Furthermore, the simulation verification process for deploying the digital twin module includes: constructing a topographic model based on a high-precision digital elevation model of the deployment area, geological survey data, and historical meteorological data. This model can accurately reflect the topographic features of the deployment area, such as altitude, slope, and land cover type.

[0030] A physical model of the equipment is constructed based on the 3D CAD model, material properties, and dynamic parameters of the new energy facility. This model simulates the physical characteristics of the equipment, such as size, weight, and power output. Finally, an operational logic model is constructed based on the system control logic and operating strategy. This model simulates the equipment's operation process, including startup, operation, and shutdown states. During the simulation, the optimal deployment scheme is transformed into a resolvable simulation script in the digital twin environment. The installation location, installation sequence, connection relationships, and control parameters of the equipment are defined in detail, and the script is injected into the digital twin model through an application programming interface (API). The simulation environment concurrently executes multiple simulation experiments, simulating different environmental conditions and operating scenarios, such as normal operating conditions, extreme weather conditions, and equipment failures, to comprehensively evaluate the adaptability of the deployment scheme. Key performance data collected during the simulation, such as energy output power, equipment operating status, and system efficiency, will be used for subsequent analysis and verification.

[0031] Furthermore, the deployed digital twin module also performs a global sensitivity analysis including the following steps: Determine the input parameters in the digital twin model, including equipment installation location deviation, ambient temperature variation range, wind speed fluctuation characteristics, solar radiation intensity and load power variation pattern, and specify a reasonable value range and probability distribution for each parameter; The Sobol sequence sampling method generates a low-bias set of sampling points in the input parameter space. This method improves the accuracy of sensitivity analysis by ensuring the uniform distribution of sampling points in the parameter space. The digital twin model is run with each sampling point as input, and the corresponding system output response is recorded, including total power generation, system efficiency, equipment load rate and operation stability indicators. The first-order sensitivity index and the total sensitivity index of each input parameter are calculated based on the principle of variance decomposition. The first-order sensitivity index measures the independent contribution of a single parameter to the output response, while the total sensitivity index measures the total contribution of the parameter itself and its interaction with other parameters to the output response. All input parameters are sorted according to their sensitivity index to identify the key parameters that have the most significant impact on system performance, providing a focus for optimizing deployment plans and making on-site adjustments.

[0032] Global sensitivity analysis determines the input parameters, their value ranges, and probability distributions in the digital twin model. The selection of input parameters needs to cover various aspects that may affect system performance. For example, deviations in equipment installation locations may affect the connection efficiency between devices; the range of ambient temperature variations and wind speed fluctuations may affect the power generation efficiency of renewable energy facilities; and solar radiation intensity and load power variation patterns directly affect the system's output power and stability.

[0033] The parameter value range and probability distribution need to be reasonably set according to the actual deployment environment and equipment characteristics to ensure the accuracy of the analysis results. Secondly, a low-biased set of sampling points is generated using the Sobol sequence sampling method. By inputting these sampling points into the digital twin model and recording the corresponding system output response, the impact of each input parameter on system performance can be comprehensively evaluated. Finally, based on the principle of variance decomposition, the first-order sensitivity exponent and the total sensitivity exponent for each input parameter are calculated. The first-order sensitivity exponent reflects the independent contribution of a single parameter to the output response, while the total sensitivity exponent considers the interaction between parameters, enabling a more comprehensive assessment of the combined impact of parameters on system performance.

[0034] Furthermore, the deployment execution control module's workflow includes: parsing the deployment plan, after module verification, into a sequence of control instructions. Each control instruction clearly defines the action to be performed, the target location, the action parameters, and the completion criteria. These control instructions cover specific operations such as equipment handling, location adjustment, connection and installation, and parameter configuration, ensuring the precise execution of the deployment process.

[0035] Based on current system state data, the module uses a predictive model to estimate the future evolution of the system state. The predictive model comprehensively considers equipment dynamics, environmental disturbances, and actuator response delays, and predicts the system's behavior under different conditions through mathematical modeling and data analysis.

[0036] The module solves a rolling time-domain optimization problem, aiming to minimize the deviation between the predicted output and the desired trajectory, while simultaneously minimizing control energy consumption. During optimization, constraints include control magnitude, control increment, and output magnitude, ensuring that control commands are effectively executed within physical and safety limits. In this way, the module can adjust control commands in real time to adapt to dynamically changing environments.

[0037] Equipment handling instructions need to clearly define the starting location, destination location, and handling route; installation instructions need to define the installation sequence, connection method, and parameter configuration. The generation of these instructions needs to take into account the physical characteristics of the equipment and the specific requirements of the deployment environment.

[0038] The construction of the predictive model requires input of current system state data, including equipment location, operating status, and environmental parameters. The model predicts the future state of the system using physical equations and data-driven methods, predicting the trajectory of the equipment using a dynamic model and forecasting the impact of external disturbances using environmental sensor data. Solving the rolling time-domain optimization problem requires setting specific optimization objectives and constraints. Optimization objectives could include the accuracy of equipment installation location and minimizing installation time, while constraints could include the maximum load capacity of the equipment and the measurement accuracy of the sensors.

[0039] Furthermore, the deployment execution control module also performs: Multidimensional time-series data is collected by deploying visual, mechanical, and position sensors on-site. Spatial features are extracted from image data using convolutional neural networks, and temporal-dependent features are extracted from time-series data using long short-term memory networks. The extracted spatial and temporal features are fused and input into a fully connected neural network for anomaly identification. Anomaly types include equipment positioning deviation exceeding limits, installation connection failure, actuator jamming, and excessive environmental interference. When an anomaly is detected, a root cause analysis process based on graph neural networks is initiated. By constructing a device correlation graph and a fault propagation model, the root cause and scope of impact of the anomaly are located. Anomaly handling strategies are generated based on deep reinforcement learning algorithms. The agent selects the optimal handling action according to the current system state and anomaly type, including adjusting installation parameters, switching to backup equipment, activating security protection mechanisms, and notifying human intervention. Establish an exception handling knowledge base to record the event description, handling measures, and handling effects of each exception, and provide handling references for similar exceptions in the future through case reasoning methods.

[0040] Specifically, multidimensional time-series data is collected through visual sensors, force sensors, and position sensors deployed on-site. Visual sensors acquire image data of the equipment's installation status, force sensors monitor the stress on the equipment, and position sensors track the equipment's installation position in real time. After preprocessing, the data collected by these sensors is fed into convolutional neural networks (CNNs) and long short-term memory networks (LSTMs), respectively.

[0041] CNNs are used to extract spatial features from image data, such as the shape, position, and connectivity of devices; LSTMs are used to extract temporal-dependent features from time-series data, such as the motion trajectory and force changes of devices. Next, the extracted spatial and temporal features are fused and input into a fully connected neural network for anomaly detection. By learning the relationships between combinations of features, the fully connected neural network can identify specific anomaly types.

[0042] When an anomaly is detected, the system initiates a root cause analysis process based on a graph neural network (GNN). By constructing an equipment correlation graph and a fault propagation model, the system locates the root cause and scope of impact of the anomaly. Finally, an anomaly handling strategy is generated based on a deep reinforcement learning algorithm. The agent selects the optimal action based on the current system state and anomaly type, such as adjusting installation parameters, switching to backup equipment, activating safety protection mechanisms, or notifying manual intervention. Simultaneously, the system establishes an anomaly handling knowledge base, recording the event description, handling measures, and effects of each anomaly, providing a reference for handling similar anomalies in the future.

[0043] Furthermore, time-series data of environmental parameters, including temperature, humidity, wind speed, light intensity, soil bearing capacity, and equipment vibration frequency, are continuously collected through weather stations, geological monitoring instruments, and equipment status sensors deployed on-site. These parameters reflect the real-time environmental conditions and equipment operating status at the deployment site, forming the basis for the system's dynamic adjustments. Secondly, the CUSUM algorithm is applied to process the environmental parameter time series data. By comparing the calculated decision function with preset thresholds, abrupt changes in environmental parameters are detected, enabling timely identification of abnormal environmental changes. The CUSUM algorithm uses accumulated deviation values ​​to determine whether environmental parameters have changed significantly, allowing for rapid response to environmental abrupt changes.

[0044] When a significant change in the environment is detected, the system initiates a strategy adjustment mechanism, employing a Kalman filter algorithm to estimate and predict environmental parameters. Kalman filtering improves the accuracy and reliability of environmental parameter predictions by combining system dynamics models with real-time observation data. Finally, based on the current environmental conditions and system state, a fuzzy inference system is used to assess the impact of environmental changes on the deployment process, and a Q-learning algorithm is applied to adjust deployment strategy parameters online. The Q-learning algorithm obtains reward signals through the interaction between the agent and the environment, continuously optimizing the decision-making strategy under changing environments to ensure the smooth progress of the deployment process.

[0045] The acquisition of environmental parameters requires selecting appropriate sensor types and deployment schemes based on the specific conditions of the deployment site. Weather stations can collect meteorological parameters such as temperature, humidity, and wind speed; geological monitoring instruments can collect parameters such as soil bearing capacity; and equipment status sensors are used to monitor operational status parameters such as equipment vibration frequency. Secondly, the implementation of the CUSUM algorithm requires setting reasonable thresholds and a method for calculating cumulative deviation. Thresholds can be set based on historical data and experience; when the cumulative deviation exceeds this threshold, a significant change in environmental parameters is considered. The implementation of the Kalman filter algorithm requires constructing an accurate system dynamics model and updating the state based on real-time observation data. The system dynamics model can be modeled based on the physical characteristics of the equipment and the laws of environmental change, and state estimation and prediction are performed using the recursive formula of the Kalman filter.

[0046] Implementing the Q-learning algorithm requires designing a reasonable reward function and state space. The reward function can be designed based on environmental adaptability; for example, a positive reward is given for successfully adapting to environmental changes, while a negative reward is given for failure or delay. The agent learns the optimal policy through interaction with the environment and dynamically adjusts the deployment policy parameters to cope with environmental changes.

[0047] Furthermore, time-dimensional data, including equipment transportation time, on-site assembly time, system debugging time, and trial operation time, is extracted from the deployment process records to calculate deployment efficiency indicators, such as deployment cycle completion rate, process integration efficiency, and resource scheduling efficiency. These indicators reflect the time management and resource utilization during the deployment process. Secondly, resource-dimensional data, including equipment usage, manpower input, and material consumption, is extracted from the resource utilization records to calculate resource utilization rate indicators, such as equipment utilization rate, manpower load rate, and material loss rate. These indicators reflect resource utilization efficiency and cost control.

[0048] Quality dimension data, including system power generation, energy conversion efficiency, operational stability, and failure rate, are extracted from system operation records to calculate deployment quality indicators such as performance compliance rate, system reliability, and ease of maintenance. These indicators reflect the system's operational performance and reliability.

[0049] The weighting coefficients of each dimension indicator are determined using a combined weighting method. Subjective weights are obtained through the analytic hierarchy process (AHP) to reflect expert judgment, while objective weights are obtained through the entropy weighting method to reflect the information content of the data itself. A weighted summation model is used to calculate the comprehensive effectiveness score. The standardized values ​​of each dimension indicator are multiplied by their corresponding weights and then summed to obtain a quantitative assessment result of deployment effectiveness.

[0050] Extracting time-related data requires combining it with time nodes in the actual deployment process. For example, equipment transportation time can be obtained from logistics system records, and on-site assembly time can be recorded from construction logs. This data is used to calculate the deployment cycle completion rate, which is the ratio of actual completion time to planned completion time; process connection efficiency is evaluated by the waiting time between each process.

[0051] Extracting resource-level data requires considering the actual usage of equipment, manpower, and materials. Equipment utilization rate can be calculated as the ratio of equipment operating time to total available time; manpower load rate can be assessed as the ratio of actual working time to planned working time; and material loss rate is calculated as the difference between actual consumption and planned consumption.

[0052] Extracting quality dimension data requires combining it with system operation monitoring data. For example, system power generation and energy conversion efficiency can be obtained through power sensors and energy monitoring equipment; operational stability and failure rate can be analyzed through fault records and system operation logs. Finally, determining the weighting coefficients requires combining expert experience and data information. Subjective weights can be calculated using expert scoring and the analytic hierarchy process (AHP), while objective weights can be calculated using the entropy weighting method based on the data's dispersion. The comprehensive performance score is calculated using a weighted summation model, which multiplies the standardized values ​​of each dimension indicator by their corresponding weights and then sums the results to obtain the final performance score.

[0053] Furthermore, during the planning phase, potential improvement opportunities are identified by analyzing historical deployment performance evaluation results. Causal analysis tools, such as fishbone diagrams, are used to systematically identify key factors affecting performance, such as inefficient equipment installation, inadequate resource allocation, or insufficient environmental adaptability. Based on these analyses, specific improvement goals are set, which must meet the SMART criteria: Specific, Measurable, Achievable, Relevant, and Time-bound.

[0054] During the execution phase, the system develops a detailed improvement plan, clearly defining responsibilities, resource allocation, and timelines. To address the issue of low equipment installation efficiency, specific measures can be implemented, such as optimizing installation processes, adding automated equipment, or training installation personnel. The project management team is clearly responsible for implementation, allocating necessary resources, and setting specific completion deadlines.

[0055] During implementation, the system comprehensively records the execution status and intermediate results of improvement measures for subsequent evaluation and adjustment. Secondly, in the inspection phase, the system collects deployment process data after the improvements and compares changes in performance indicators before and after the improvements. Statistical testing methods, such as t-tests or analysis of variance, are used to verify the significance of the improvements and analyze the reasons for not achieving the expected goals, such as changes in the external environment or internal execution deviations. Finally, in the handling phase, the system incorporates proven effective improvement measures into standard operating procedures (SOPs) and promotes them throughout the organization through training and culture building. The optimized installation process is incorporated into SOPs, and internal training and culture building ensure that all relevant personnel can master and implement these improvement measures. Simultaneously, unresolved issues are carried over to the next improvement cycle, forming a closed loop of continuous improvement.

[0056] The foregoing description is illustrative of the invention and should not be construed as limiting it. Although several exemplary embodiments of the invention have been described, those skilled in the art will readily understand that many modifications can be made to the exemplary embodiments without departing from the novel teachings and advantages of the invention. Therefore, all such modifications are intended to be included within the scope of the invention as defined in the claims. It should be understood that the foregoing description is illustrative of the invention and should not be construed as limiting it to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The invention is defined by the claims and their equivalents.

Claims

1. A modular rapid deployment system for new energy facilities, characterized in that, include: The deployment planning and decision-making module is used to receive geospatial data of the deployment environment, specification parameter data of new energy facilities, and timeliness requirements data of deployment tasks. It solves the optimal deployment scheme by establishing a multi-objective optimization model and considering various practical constraints. Deploy a digital twin module to build a virtual deployment environment based on the optimal deployment scheme, including terrain features, equipment physical characteristics and operating logic, and verify the feasibility and robustness of the deployment scheme under various working conditions through simulation; The deployment execution control module is used to decompose the verified deployment scheme into a sequence of control instructions with time sequence and logical relationship, and to adaptively adjust the control parameters by monitoring the deviation between the execution status and the expected status in real time. The dynamic environment adaptation module is used to continuously collect environmental parameter data and equipment operating status data at the deployment site, identify significant changes in environmental conditions through a change point detection algorithm, and dynamically update the deployment strategy based on an online learning mechanism. The deployment performance evaluation module is used to collect data from the beginning to the end of the deployment process, establish a multi-dimensional evaluation system including efficiency indicators, quality indicators, cost indicators and reliability indicators, and generate optimization suggestions through data analysis.

2. The modular rapid deployment system for new energy facilities according to claim 1, characterized in that, The deployment planning and decision-making module executes the following steps: Collect geospatial data on the latitude and longitude coordinates, altitude, terrain slope, and land cover type of the deployment area; obtain specification parameters of the size, weight, power output characteristics, and interface standards of the new energy facilities to be deployed; and receive deployment task requirements data specified by the user, such as deployment completion time, budget limit, and performance indicators. Establish a multi-objective optimization model that simultaneously considers minimizing deployment costs, minimizing deployment time, and maximizing resource utilization. Deployment costs include equipment procurement costs, transportation costs, installation costs, and maintenance costs. Deployment time includes equipment transportation time, on-site assembly time, system debugging time, and trial operation time. Resource utilization includes land utilization rate, equipment utilization rate, and energy conversion efficiency. An improved non-dominated sorting genetic algorithm is used to solve the multi-objective optimization model. This algorithm improves search efficiency and solution quality by introducing an elite retention strategy and an adaptive crossover mutation operator, and finally outputs a set of Pareto optimal deployment schemes. Based on a decision-making method combining entropy weighting and ideal point methods, the deployment scheme that best meets the actual needs is selected from the Pareto optimal solution set. The entropy weighting method is used to objectively determine the weight of each indicator, while the ideal point method is used to evaluate the degree of closeness of each scheme to the ideal scheme.

3. The modular rapid deployment system for new energy facilities according to claim 2, characterized in that, The solution process of the improved non-dominated sorting genetic algorithm specifically includes: The Latin hypercube sampling method is used to generate an initial population with good spatial distribution characteristics within the feasible region of the decision variables, where the decision variables include equipment layout coordinates, number of equipment, connection topology and installation sequence. Based on the Pareto dominance relationship, individuals in the population are ordered into non-dominated layers, and the population is divided into multiple non-dominated layers. The first layer consists of individuals that are not dominated by any other individuals, the second layer consists of individuals that are dominated only by individuals in the first layer, and so on. The crowding distance of individuals in the same non-dominated layer is calculated. This distance is used to assess the distribution density of individuals in the target space by measuring the sum of the distances between an individual and its neighboring individuals in each objective function direction. A binary tournament selection method is used to select parent individuals from the current population. The selection criteria prioritize individuals with high non-dominant ranking, and within the same ranking, individuals with large crowding distance are selected to maintain population diversity. Simulated binary crossover and polynomial mutation operations are performed on the selected parent individuals to generate offspring individuals, wherein the crossover probability and mutation probability are adaptively adjusted according to the population iteration generation and the individual distribution. The parent and offspring populations are merged, and a new generation of population is generated through environmental selection. Environmental selection takes into account non-dominant ranking and crowding distance to ensure that superior individuals are preserved while maintaining population diversity.

4. The modular rapid deployment system for new energy facilities according to claim 1, characterized in that, The deployment of the digital twin module performs a simulation verification process including the following steps: A topographic model is constructed based on a high-precision digital elevation model of the deployment area, geological survey data, and historical meteorological data; a physical model of the equipment is constructed based on a 3D CAD model of the new energy facilities, material properties, and dynamic parameters; and an operation logic model is constructed based on system control logic and operation strategies. The optimal deployment scheme is transformed into a resolvable simulation script for the digital twin environment. This script defines in detail the device installation location, installation sequence, connection relationship and control parameters, and is injected into the digital twin model through the application programming interface. Multiple simulation experiments are executed concurrently in a digital twin environment. Each experiment simulates different environmental conditions and operating scenarios, including normal operating conditions, extreme weather conditions, and equipment failures, to comprehensively evaluate the adaptability of the deployment plan. Time series data of key system performance indicators during the simulation process are collected, including energy output power, equipment operating status, system efficiency and stability indicators, and the data are preprocessed and feature extracted. Statistical hypothesis testing methods are used to analyze simulation results and determine whether the deployment scheme meets the preset performance requirements and safety standards, including mean test, variance test and distribution fit test.

5. The modular rapid deployment system for new energy facilities according to claim 4, characterized in that, The deployment of the digital twin module also performs: Determine the input parameters in the digital twin model, including equipment installation location deviation, ambient temperature variation range, wind speed fluctuation characteristics, solar radiation intensity and load power variation pattern, and specify a reasonable value range and probability distribution for each parameter; The Sobol sequence sampling method generates a low-bias set of sampling points in the input parameter space. This method improves the accuracy of sensitivity analysis by ensuring the uniform distribution of sampling points in the parameter space. The digital twin model is run with each sampling point as input, and the corresponding system output response is recorded, including total power generation, system efficiency, equipment load rate and operation stability indicators. The first-order sensitivity index and the total sensitivity index of each input parameter are calculated based on the principle of variance decomposition. The first-order sensitivity index measures the independent contribution of a single parameter to the output response, while the total sensitivity index measures the total contribution of the parameter itself and its interaction with other parameters to the output response. All input parameters are sorted according to their sensitivity index to identify the key parameters that have the most significant impact on system performance, providing a focus for optimizing deployment plans and making on-site adjustments.

6. The modular rapid deployment system for new energy facilities according to claim 1, characterized in that, The deployment execution control module performs the following steps: The verified deployment scheme is parsed into a sequence of control instructions with strict time order and logical dependencies. Each control instruction clearly specifies the action to be performed, the target location, the action parameters, and the completion criteria. At the beginning of each control cycle, based on the currently collected system state data, a prediction model is used to estimate the evolution of the system state over a future period of time. The prediction model comprehensively considers the dynamic characteristics of the equipment, environmental disturbance factors, and actuator response delay. Solve the rolling time-domain optimization problem with multiple constraints. The goal is to minimize the deviation between the predicted output and the desired trajectory while minimizing the control energy consumption. The constraints include control magnitude constraints, control increment constraints, and output magnitude constraints. The first control quantity in the control command sequence obtained by optimization is sent to the actuator to control the field equipment to complete the specified actions, including equipment handling, position adjustment, connection and installation, and parameter configuration. The actual output response of the system is collected in real time by a sensor network. The deviation between the actual response and the predicted response is calculated. This deviation is used to correct the state estimation of the prediction model and enhance the robustness of the model in environments with uncertainty.

7. The modular rapid deployment system for new energy facilities according to claim 6, characterized in that, The deployment execution control module also performs the following steps: Multidimensional time-series data is collected by deploying visual, mechanical, and position sensors on-site. Spatial features are extracted from image data using convolutional neural networks, and temporal-dependent features are extracted from time-series data using long short-term memory networks. The extracted spatial and temporal features are fused and input into a fully connected neural network for anomaly identification. Anomaly types include equipment positioning deviation exceeding limits, installation connection failure, actuator jamming, and excessive environmental interference. When an anomaly is detected, a root cause analysis process based on graph neural networks is initiated. By constructing an equipment correlation graph and a fault propagation model, the root cause and scope of impact of the anomaly are located. Anomaly handling strategies are generated based on deep reinforcement learning algorithms. The agent selects the optimal handling action according to the current system state and anomaly type, including adjusting installation parameters, switching to backup equipment, activating security protection mechanisms, and notifying human intervention. Establish an exception handling knowledge base to record the event description, handling measures, and handling effects of each exception, and provide handling references for similar exceptions in the future through case reasoning methods.

8. The modular rapid deployment system for new energy facilities according to claim 1, characterized in that, The dynamic environment adaptation module performs the following steps: Time-series data of environmental parameters, including temperature, humidity, wind speed, light intensity, soil bearing capacity, and equipment vibration frequency, are continuously collected through weather stations, geological monitoring instruments, and equipment status sensors deployed on site. The cumulative sum algorithm is applied to process the time series of environmental parameters. By comparing the calculated decision function with the preset threshold, abrupt changes in environmental parameters are detected, and abnormal environmental changes are discovered in a timely manner. The Kalman filter algorithm is used to estimate and predict environmental parameters. By combining the system dynamics model and real-time observation data, the accuracy and reliability of environmental parameter prediction are improved. When a significant change in the environment is detected, a strategy adjustment mechanism is activated. Based on the current environmental conditions and system state, a fuzzy inference system is used to assess the impact of the environmental change on the deployment process. The Q-learning algorithm is applied to adjust deployment strategy parameters online. The agent obtains reward signals by interacting with the environment and continuously optimizes its decision-making strategy in changing environments to ensure the smooth progress of the deployment process.

9. The modular rapid deployment system for new energy facilities according to claim 1, characterized in that, The deployment performance evaluation module performs the following steps: Extract time-dimensional data from the deployment process records, including equipment transportation time, on-site assembly time, system debugging time, and trial operation time, and calculate deployment efficiency indicators such as deployment cycle completion rate, process connection efficiency, and resource scheduling efficiency; Extract resource dimension data from resource utilization records, including equipment usage, manpower input, and material consumption, and calculate resource utilization indicators such as equipment utilization rate, manpower load rate, and material loss rate. Extract quality dimension data from system operation records, including system power generation, energy conversion efficiency, operational stability and failure rate, and calculate deployment quality indicators such as performance compliance rate, system reliability and maintenance convenience; The weight coefficients of each dimension index are determined by a combined weighting method. The subjective weights are obtained through the analytic hierarchy process to reflect expert experience and judgment, while the objective weights are obtained through the entropy weighting method to reflect the information content of the data itself. The comprehensive performance score is calculated by multiplying the standardized values ​​of each dimension index by their corresponding weights and then summing the results to obtain a quantitative assessment of deployment performance. Based on the evaluation results, specific improvement suggestions are generated, and optimization measures are proposed for the less efficient links, including process reengineering, resource allocation optimization, and technical solution improvement.

10. The modular rapid deployment system for new energy facilities according to claim 9, characterized in that, The deployment performance evaluation module also performs the following steps: During the planning phase, improvement opportunities are identified based on historical deployment effectiveness evaluation results. Causal analysis tools are used to identify key factors affecting effectiveness, and specific, measurable, achievable, relevant, and time-bound improvement goals are set. During the implementation phase, a detailed implementation plan for improvement measures should be developed, clearly defining the division of responsibilities, resource allocation, and timelines. The implementation status and intermediate results of the improvement measures should be fully recorded during the implementation process. During the inspection phase, data on the improved deployment process are collected, the changes in performance indicators before and after the improvement are compared, statistical testing methods are used to verify the significance of the improvement effect, and the reasons for not achieving the expected goals are analyzed. During the handling phase, proven effective improvement measures are incorporated into standard operating procedures and promoted throughout the organization through training and culture building. Meanwhile, unresolved issues are carried over to the next improvement cycle. Establish a quantitative evaluation model for improvement effects. This model comprehensively considers the degree of improvement in deployment efficiency, the degree of improvement in deployment quality, the degree of cost reduction, and the degree of reliability improvement, and calculates the comprehensive score of improvement effects by weighted summation.

Images