Power supply method for factory area

Abstract

The present application relates to a plant power supply method, aiming to solve the problems of inaccurate sampling, lack of quantitative support for loop grouping and low digitalization degree of construction and operation in existing plant power supply planning. The method comprises: binding the device identifier with the physical space position, establishing an initial digital twin base map containing three-dimensional coordinates and power parameters; collecting the power consumption characteristic vectors of the device under each steady state through the full-gear excitation protocol, and constructing a discrete characteristic state set; comprehensively considering the electrical characteristic similarity and spatial proximity for intelligent grouping of loops, and pre-preparing abnormal detection rules; generating digital work orders to guide construction, and delivering digital twins carrying characteristic fingerprints and diagnostic logic. Through the above scheme, the present application realizes accurate sampling of power supply parameters and scientific planning of loops, effectively reduces loop interference and resource waste, constructs a digital management system with real-time early warning capability, and significantly improves the stability and whole life cycle management efficiency of the plant power supply system.

Description

Technical Field

[0001] This invention relates to power systems, and more particularly to methods for power supply in industrial plants. Background Technology

[0002] As modern industrial plants undergo increasingly intelligent and digital transformation, the power supply system, as a core infrastructure ensuring the stable operation of various production equipment, is becoming increasingly important. Industrial production's reliance on electricity supply continues to rise, and the rationality of the power supply system's layout and operational stability directly affect the plant's production efficiency, energy economy, and overall operational safety.

[0003] The planning and construction of power supply circuits for plant equipment is a crucial aspect of power supply system construction, involving multiple technical dimensions such as accurate sampling of equipment power consumption characteristics, logical design of circuit grouping, scientific selection of hardware specifications, and digital delivery. In-depth analysis of equipment power consumption characteristics and intelligent planning of circuit structures can effectively improve the scientific nature of power distribution, providing solid data support for subsequent intelligent operation and maintenance and abnormal state early warning.

[0004] The existing technology for planning and constructing power supply circuits in factory areas suffers from the following main problems. First, the sampling phase lacks scientific planning strategies and standardized processes, relying heavily on manual experience for static sampling. This fails to comprehensively cover the power consumption characteristic vectors of equipment under different operating conditions, resulting in low accuracy and insufficient representativeness of the collected data. Second, the circuit planning phase lacks quantitative analysis and intelligent algorithm support. Grouping logic often only considers spatial proximity while ignoring the similarity of electrical characteristics, easily causing current fluctuations and mutual interference within the circuit. Furthermore, the selection of cable and distributor specifications is often arbitrary, making it difficult to achieve precise matching between resource investment and power demand. Third, the level of digital control during construction is low, relying heavily on paper-based work orders. This makes it difficult to guarantee construction accuracy, and the delivered results lack digital archives containing equipment fingerprints and anomaly detection rules, leading to low efficiency in troubleshooting and a lack of proactive early warning capabilities during later maintenance. Finally, the lack of a standardized binding mechanism between equipment identification, physical location, and sampling data makes it easy for the planning, sampling, and construction stages to become disconnected, making it difficult to build a digital twin that combines physical topology description and real-time diagnostic capabilities. These problems collectively restrict the intelligent upgrading and full life cycle management efficiency of power supply systems in industrial plants. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a power supply method for power plants to overcome the above-mentioned defects in the existing technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The power supply method for the factory area includes the following steps: Step 1, Equipment and Sampling Point Association Steps: Read the factory layout drawings through the central processing server, analyze and extract the building axes, equipment installation reference points and cable tray paths, and establish the factory coordinate system; use identification technology to bind the physical equipment on site with the equipment identification on the drawings, obtain the three-dimensional coordinates of the equipment through the spatial positioning system, and establish a file containing identification code, equipment type, location and rated power for each piece of equipment to be powered in the initial digital twin base map; Step 2, Discrete Feature Sampling Step: Based on the three-dimensional coordinates of the equipment, the optimal temporary sampling points are planned using a cluster center iterative strategy; the full-range excitation protocol is executed through a mobile sampling system to control the equipment to traverse standby, half-load, and full-load states; the real-time fluctuation rate of power consumption characteristics is monitored, and after steady state, power consumption feature vectors such as active power and reactive power are collected to construct a set of discrete feature states of the equipment; Step 3, Intelligent Planning Steps: Define a similarity function for comprehensive electrical feature similarity and spatial proximity, and use a clustering algorithm to group loops; extract peak electrical parameters to specify distributor specifications, and combine maximum steady-state current, power supply distance and measured data to calibrate line loss and determine cable specifications; combine discrete feature states of equipment to form a complete set of legal steady-state features and pre-form anomaly detection rules; Step 4, Digital Construction and Delivery: Generate digital work orders based on the plan, compare deviations after construction is completed, and form a digital twin of the power supply system that carries discrete feature fingerprints and anomaly detection rules.

[0007] Preferably, the identification binding process in step one includes: obtaining the unique identification code of the device using radio frequency identification or QR code scanning, and mapping and associating it with the virtual device object; the spatial positioning system combines laser scanning point cloud data with the drawing for fitting and calibration to ensure that the position deviation is within a preset range, and constructing an initial digital twin base map with spatial topological relationships.

[0008] Preferably, the full-range excitation and steady-state determination in step two includes: the mobile sampling system sends a command to switch the typical working range of the equipment, monitors the power consumption characteristics through a high-precision current transformer and calculates the rolling standard deviation as the volatility; when the volatility is lower than the first preset threshold and the duration reaches the target, it is determined that the system has entered a steady state and triggers the acquisition of feature snapshots. The power consumption feature vector also includes the phase angle and amplitude of specific order current harmonics.

[0009] Preferably, the optimal temporary sampling point planning in step two includes: dynamically clustering based on the three-dimensional coordinates of the device, minimizing the sum of squares of the lengths of temporary cables within the cluster, solving for the optimal sampling points using the gradient descent algorithm, and outputting a set of sampling points to reduce the impact of temporary cable impedance on sampling accuracy.

[0010] Preferably, the loop grouping in step three includes: assigning weights to electrical characteristic similarity and spatial proximity, using agglomerative hierarchical clustering algorithm to group devices with similar characteristics and close locations into the same loop; and combining harmonic phase characteristics to group devices with opposite harmonic phases to reduce interference by utilizing the harmonic self-cancellation effect.

[0011] Preferably, the electrical feature similarity calculation includes: finding the optimal matching feature pair of the discrete feature state sets of two devices, calculating the weighted Euclidean distance mean as the optimal matching distance; quantifying the differences in the electrical characteristics of the devices, and classifying them into the same circuit if the optimal matching distance is less than a preset threshold.

[0012] Preferably, the specification of the distributor in step three includes: extracting the maximum instantaneous current and starting power of a subset of the same distributor equipment, ensuring that the rated current of the distributor covers the sum of the maximum demand and combining the power consumption simultaneity rate and the dynamic redundancy coefficient, with the redundancy coefficient adjusted according to the power consumption simultaneity rate range.

[0013] Preferably, the cable specification calculation in the above steps includes: calculating the theoretical voltage drop based on the maximum steady-state current and power supply distance, deriving the actual line loss coefficient and calibrating the model based on measured data; and correcting the equivalent resistance based on the ambient temperature and cable temperature rise, and selecting a cable cross-sectional area that meets the requirements of current carrying capacity and voltage drop.

[0014] The factory's power supply system includes: The central processing server executes logic such as data parsing and coordinate transformation, and deploys digital twin base map management modules, electrical feature databases, etc., to establish the plant area coordinate system and initial digital twin base map; The mobile sampling system, including an autonomous mobile chassis and a power quality analyzer, executes a discrete feature sampling and excitation protocol to collect and upload power consumption feature vectors. On-site sensing terminals, including identification tags and spatial positioning terminals, establish a mapping between physical and digital spaces; The digital construction guidance terminal uses an augmented reality industrial tablet computer to achieve visual guidance of construction. The intelligent planning engine generates loop grouping and cable specification schemes, and the anomaly detection rule generator constructs a complete set of legal steady-state features and embeds them into the digital twin. The digital twin monitors loop parameters in real time, and triggers alarms and highlights them when anomalies occur.

[0015] The beneficial effects of this invention are as follows: By executing a full-range excitation protocol and discrete feature sampling, this invention can obtain real power consumption data of equipment under all operating conditions, completely solving the specification mismatch problem caused by relying on rated parameters or manual experience in traditional planning. Cable specification calculation based on measured line loss calibration enables cable selection to optimize material costs while meeting safety requirements, effectively avoiding resource waste. This invention defines a similarity function that comprehensively considers electrical characteristics and spatial distance, and uses clustering analysis algorithms for loop division. This method not only shortens wiring length, but more importantly, by classifying equipment with similar electrical characteristics, it reduces electromagnetic interference between different types of loads, improving the overall power quality and equipment operational stability of the plant's power supply system. Attached Figure Description

[0016] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a flowchart illustrating the association between the device and the sampling point in this invention. Detailed Implementation

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

[0018] It should be noted that when a component is described as "fixed to" another component, it can be directly on the other component or may have a component in between. When a component is considered "connected to" another component, it can be directly connected to the other component or may have a component in between. When a component is considered "set on" another component, it can be directly set on the other component or may have a component in between. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0020] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings: A power supply method for factory areas is proposed. This method achieves closed-loop management of the entire process, from factory planning, equipment feature acquisition, intelligent circuit design to digital construction and delivery, by constructing a highly integrated hardware system and digital twin logic framework.

[0021] At the system architecture level, this embodiment constructs a hardware integrated system consisting of a central processing server, a mobile sampling system, on-site sensing terminals, and digital construction guidance terminals. The central processing server, as the core brain of the system, is equipped with a high-performance central processing unit, a large-capacity solid-state storage array, and a dedicated graphics processing unit for running clustering analysis algorithms. Internally, the server deploys a digital twin base map management module, an electrical feature database, an intelligent planning engine, and an anomaly detection rule generator. The central processing server is connected to the plant's backbone network via industrial-grade 10 Gigabit Ethernet, ensuring high-speed transmission of massive amounts of sampling data and model information.

[0022] The mobile sampling system is the key physical entity for acquiring the underlying data in this system. Its structure includes a chassis with autonomous mobility, on which integrates a high-precision power quality analyzer, a programmable logic controller (PLC), a high-power electronic load, a multi-channel signal switching relay group, and an industrial wireless communication module. The power quality analyzer's current acquisition end uses a high-sensitivity Rogowski coil or a high-precision current transformer, capable of capturing current fluctuations from milliamperes to kiloamperes. Its sampling frequency is set to 20 kHz to meet the requirements for accurate analysis of higher harmonics. The PLC is responsible for executing the full-range excitation protocol, controlling the power supply and range switching of the device under test (DUT) through the relay group.

[0023] The on-site sensing terminals include RFID tags or QR code labels deployed on the equipment side, as well as handheld laser scanners and spatial positioning terminals carried by construction workers. These hardware components together support the coordinate mapping relationship between physical space and virtual digital space. The digital construction guidance terminal uses an industrial tablet computer with augmented reality display capabilities, which, through an integrated 9-axis gyroscope and high-pixel camera, realizes a 3D visualization of the construction path and installation location.

[0024] Supported by the above system architecture, the power supply method for the factory area described in this embodiment operates according to the following dynamic logic: The first step involves associating equipment with sampling points. The central processing server, through its data import interface, reads vectorized CAD drawings or high-precision BIM models of the factory buildings. The digital twin base map management module parses the vector elements in the drawings, automatically extracting the building's walls, columns, cable trays, and pre-set equipment installation points, and converting them into digital entities with three-dimensional geographic coordinates. Construction personnel use handheld laser scanners to scan the factory site, acquiring point cloud data of the physical space, and then fitting and calibrating it with the imported vector drawings to establish a unified factory coordinate system. Subsequently, using RFID technology, construction personnel bind RFID tags on physical equipment to virtual equipment identifiers in the digital twin base map. Each time a binding is completed, the spatial positioning terminal records the current 3D spatial coordinates of the equipment in real time and sends this coordinate information back to the central processing server, thus establishing a complete file for each piece of equipment to be powered in the initial digital twin base map, including its identification code, equipment type, physical location, and rated power.

[0025] The discrete feature sampling step then begins. The central processing server invokes the optimal sampling point planning algorithm based on the bound device location information. This algorithm employs a cluster-center iterative strategy. First, devices within a 15-meter physical distance are divided into sampling clusters. Then, a geometric center position within each cluster is found as the optimal temporary sampling point. This ensures that when the mobile sampling system is deployed at this point, the total length of temporary cables connecting to all devices within the cluster is minimized, effectively suppressing measurement errors caused by cable distributed impedance. The mobile sampling system moves to the designated sampling point according to the planning instructions, and its internal programmable logic controller establishes a control connection with the device under test via an industrial wireless communication module.

[0026] During sampling, the system executes a preset full-range excitation protocol. The active control module of the mobile sampling system sends a specific control sequence to the device under test (DUT), forcing the device to traverse all its typical operating states, such as standby, startup, no-load, half-load, and full-load, in order from low load to high load. The power quality analyzer monitors the voltage and current waveforms in the circuit in real time and converts them into feature vectors such as active power, reactive power, current harmonic distortion rate, and power factor. To ensure the representativeness of the feature data, the system introduces steady-state determination logic: at each operating range, the system continuously calculates the rolling standard deviation of the feature vectors. When this standard deviation remains below 0.5% of the mean for 30 consecutive seconds, the device is determined to have entered a steady state. At this time, the system triggers a high-precision snapshot acquisition to record the feature vectors under this steady state. Finally, a discrete feature state set consisting of multiple steady-state feature vectors is generated for each device and stored in the electrical feature database of the central processing server.

[0027] After entering the intelligent planning step, the intelligent planning engine of the central processing server takes over the data. The engine first defines a similarity function between devices, which is a weighted model that integrates electrical similarity and spatial distance. When calculating electrical similarity, the engine extracts the harmonic distribution matrix from the set of discrete feature states of two devices, and quantifies the difference in their power consumption characteristics by calculating the weighted Euclidean distance between the two matrices. The spatial distance is calculated based on the 3-dimensional coordinates in the digital twin base map. The intelligent planning engine adopts an agglomerative hierarchical clustering algorithm to automatically group devices with similar electrical characteristics and adjacent physical locations into the same power supply circuit.

[0028] For the divided circuits, the system automatically specifies the specifications of sockets or distributors. This process follows the coverage redundancy logic, that is, the system traverses the maximum instantaneous current of all devices in the circuit at each gear, and calculates the possible combined peak value. Taking into account the simultaneous operation rate of equipment in industrial production, the system automatically adjusts the redundancy coefficient according to the preset production schedule. If the devices in this circuit have strong synchronization in the process flow, the redundancy coefficient is set to 1.25; otherwise, it is set to 1.1.

[0029] In the cable specification calculation link, the system adopts a measured line loss calibration analysis strategy. The central processing server first calculates the theoretical voltage drop according to the physical length of the circuit and the maximum steady-state current, and then extracts the measured voltage drop data recorded by the mobile sampling system in the sampling stage when a temporary cable of a specific length is used. By comparing the theoretical value with the measured value, the system reversely derives the actual line loss coefficient of the cable in this environment, and revises the selection scheme of cable cross-sectional area accordingly, so as to ensure that the lowest cost cable specification is selected on the premise of meeting the voltage drop requirement of less than 5%.

[0030] Finally comes the digital construction and delivery step. The central processing server converts the planning results into a digital work order containing 3-dimensional path coordinates, cable models, wiring terminal definitions and hardware installation lists. The digital construction guidance terminal uses augmented reality technology to overlay the virtual wiring path and installation instructions on the real-time image of the construction site. After the construction personnel complete the installation, they use the terminal's camera to take photos of the construction results and re-measure the spatial coordinates. The system automatically compares the re-measured data with the planning model, and if the deviation is within 5 cm, the construction is judged as qualified.

[0031] During the delivery phase, the system generates a digital twin of the power supply system, containing discrete feature fingerprints of all devices and pre-defined anomaly detection rules. This digital twin embeds analytical logic based on a state-matching frequency monitoring strategy. In subsequent operation, when the real-time steady-state feature vector collected by the on-site intelligent distribution cabinet deviates from the complete set of valid steady-state feature states stored in the digital twin, and the deviation exceeds a 10% threshold, the digital twin will highlight the faulty circuit in the virtual space and automatically push a diagnostic report containing the cause of the fault, affected equipment, and maintenance recommendations.

[0032] In terms of hardware configuration, the mobile sampling system in this embodiment adds an active filter module and a high-frequency pulse interference simulator. The active filter module is used to eliminate background noise from the power grid upstream of the sampling point during the sampling process, ensuring that the collected equipment characteristics are only related to the nonlinear load characteristics of the equipment itself. The high-frequency pulse interference simulator is used to inject interference signals of a specific frequency into the sampling loop under controlled conditions to test the robustness characteristics of the equipment under abnormal power environments, and to record the interference signals as auxiliary feature vectors into the discrete feature state set.

[0033] In performing the discrete feature sampling step, this embodiment employs a deeper data white-boxing process. The high-precision current transformer in the sampling system converts the current signal into a 16-bit binary two's complement digital stream, which is then subjected to real-time Fast Fourier Transform via a Field-Programmable Gate Array (FPGA). The FPGA not only extracts conventional active and reactive power parameters but also performs refined decomposition of the phase angles and amplitudes of the 1st to 50th harmonics. Each steady-state eigenvector is expanded into a hyperspace vector containing more than 100 dimensions.

[0034] To address these high-dimensional feature vectors, the similarity calculation logic in the intelligent planning step has been upgraded. This embodiment introduces a manifold learning algorithm to project high-dimensional electrical feature vectors into a low-dimensional manifold space, revealing potential electrical correlations between different devices. When planning loop grouping, the system not only considers the superposition of current amplitudes but also utilizes the phase cancellation principle to attempt to group devices with opposite harmonic phases into the same loop, thereby improving power quality through the self-cancellation effect of harmonics within the loop.

[0035] Specifically, the calculation of electrical feature similarity adopts the following formulaic strategy: For devices A and B, their discrete characteristic state sets are set M and set N, respectively. The system constructs a matching matrix by finding the minimum weighted distance between element mi in set M and element nj in set N. This distance calculation considers not only the difference in current amplitude but also assigns higher weight coefficients to characteristic harmonics such as the 3rd, 5th, and 7th harmonics. Finally, by calculating the ratio of the trace of the entire matching matrix to the total number of elements, the discrete distribution difference index of the overall power consumption characteristics of the two devices is obtained.

[0036] In the calibration phase of cable specification calculation, this embodiment adds a correction model based on temperature field coupling. The mobile sampling system simultaneously records ambient temperature and cable surface temperature rise data during sampling. The physical simulation engine in the central processing server dynamically corrects the equivalent resistance under different load levels based on these measured temperature rise data, combined with the skin effect and proximity effect of the cable. This process enables the generated cable specification scheme to better address heat dissipation problems caused by high temperatures or dense wiring in factory areas during summer, further improving the safety of the power supply system.

[0037] In the digital construction phase, this embodiment introduces an automated cabling robot based on LiDAR positioning. The robot reads the path information from the digital work order and uses its built-in robotic arm to lay cables and fix supports. A central processing server receives the construction trajectory data uploaded by the robot in real time and compares it with the digital twin map. This machine-collaborative construction mode controls the positional deviation to within 2 centimeters, greatly improving construction accuracy.

[0038] The delivered digital twin exhibits enhanced intelligence during the operation and maintenance phase. Because it internally stores high-dimensional feature fingerprints of equipment under different operating conditions, when latent faults such as insulation aging, poor connector contact, or internal component wear occur in the circuit, these faults subtly alter the circuit's overall steady-state feature vector. The analysis engine within the digital twin compares the real-time vector with the complete set of legitimate states in the fingerprint database, using a support vector machine algorithm for classification and identification. This allows for precise determination of which device has experienced what type of potential fault, thus achieving a leap from "reactive maintenance" to "predictive maintenance."

[0039] At the system architecture level, this embodiment adopts a distributed processing architecture. The central processing server acts as the overall control node, while each floor of the plant deploys a regional edge computing gateway. These edge computing gateways are responsible for processing the raw waveform data uploaded by the mobile sampling systems within their respective floors, performing preliminary feature extraction and data cleaning, and only uploading the compressed feature vectors to the central server. This architecture effectively alleviates the bandwidth pressure on the plant's backbone network during large-scale sampling.

[0040] In the process of associating devices with sampling points in step 1, the system particularly enhances its ability to process the 3D spatial topology of multi-story buildings. The digital twin base map management module not only records the planar coordinates of the devices, but also obtains the floor information and absolute altitude of the devices through barometric altimeters and UWB positioning technology. When establishing the initial digital twin base map, the system automatically identifies the locations of cable shafts spanning multiple floors and uses them as key constraint nodes in loop planning.

[0041] In the discrete feature sampling of step 2, considering the large number of devices in the electronics manufacturing plant, the system adopts a multi-machine collaborative sampling strategy. Multiple mobile sampling systems work in parallel within the same area, and the central processing server dynamically allocates tasks to each sampling system in real time using a clustering center iterative algorithm. When solving for the optimal sampling point location, the algorithm introduces a penalty term to avoid cross-interference caused by temporary wiring between different sampling systems.

[0042] The excitation protocol within the sampling system has also been optimized for electronic devices. Considering that electronic devices typically have complex switching power supply structures, with significant fluctuations in startup current and standby power, the system incorporates frequency domain stability verification into its steady-state determination logic. Feature vector acquisition is only triggered when the current spectrum distribution remains constant for 20 consecutive seconds and the fundamental frequency fluctuation is less than 0.1 Hz. This rigorous determination mechanism ensures that the acquired feature data accurately reflects the true load characteristics of the electronic device under precise manufacturing conditions.

[0043] In the intelligent planning stage of step 3, the system introduces dynamic voltage drop simulation. Because electronic manufacturing equipment is extremely sensitive to voltage fluctuations, the intelligent planning engine simulates transient voltage fluctuations when all devices in the loop start or stop simultaneously during cable specification calculations. The system utilizes the startup transient waveform characteristics acquired in step 2 to predict the voltage drop depth of the loop bus through convolution operations. If the simulation results show that the voltage drop exceeds the equipment's allowable range, the system will automatically increase the cable cross-sectional area or add a local energy storage / filtering compensation device to the planning, and update the digital work order.

[0044] In the digital construction and delivery process, this embodiment emphasizes a closed-loop quality control system for the construction process. The digital work order includes requirements for digital recording of wiring torque and terminal crimping quality. Construction personnel use digital torque wrenches with Bluetooth communication capabilities for wiring; torque data is uploaded in real-time to the digital construction guidance terminal and linked to the equipment identifier.

[0045] The resulting digital twin of the power supply system not only includes the electrical topology but also integrates the plant's thermal distribution map. By reading current data from the circuits, the digital twin uses a built-in thermodynamic model to calculate the heat generation of each cable segment and distributor in real time, and combines this with the operating status of the plant's air conditioning system to assess the thermal safety of the power supply system. When the risk of heat accumulation in a localized area is high, the digital twin will proactively issue an early warning to maintenance personnel, suggesting adjustments to production schedules or optimization of ventilation conditions.

[0046] In terms of system hardware, the mobile sampling system is equipped with industrial isolation transformers and special high-current transformers that are resistant to high voltage and high current surges. The chassis of the sampling system adopts an enhanced shock-resistant structure to cope with ground vibrations in machining sites. The central processing server enhances its ability to process nonlinear, high-load fluctuation data and is configured with a dedicated real-time stream processing framework.

[0047] In Step 1, the system connects to the factory's MES system to obtain the equipment's processing parameters. These parameters are injected as metadata into the digital twin base map, enabling the system to understand the differences in power consumption requirements of the equipment under different processing steps.

[0048] In the discrete feature sampling of step 2, the full-range excitation protocol is extended to a "process flow excitation protocol." The mobile sampling system not only controls the electrical range of the equipment but also simulates load changes during actual processing. For example, by controlling the electronic load to simulate the resistance changes during tool cutting, it collects dynamic power consumption feature vectors that most closely resemble the real production scenario. The data processing module of the sampling system uses wavelet transform technology to analyze the non-stationary current signal and extract the characteristic frequencies reflecting mechanical wear and load impact.

[0049] In the intelligent planning logic of step 3, the system focuses on addressing the impact of high-power impulsive loads on circuit stability. A "power fluctuation complementarity" dimension has been added to the similarity function. The system attempts to group components with complementary power consumption characteristics into the same circuit to smooth out the total current fluctuations in the circuit. When specifying distributor specifications, the system employs a probability density-based redundancy calculation model, using Monte Carlo simulation to predict the maximum current probability distribution of the circuit under various operating condition combinations. This maximizes hardware specification redundancy and reduces construction costs while ensuring safety.

[0050] In the cable specification calculation, the system specifically considers the inductive voltage drop caused by long-distance power transmission in heavy industrial areas. Based on the complex impedance parameters measured in step 2, the system constructs an accurate transmission line model in digital space and determines the optimal cable cross-sectional area and laying method (such as triangular arrangement or horizontal arrangement) through iterative calculation to reduce the electromagnetic inductance between cables.

[0051] In step 4, the delivery phase, the digital twin is endowed with enhanced fault backtracking capabilities. Since heavy equipment failures are often destructive, the digital twin establishes a "black box" function based on a time-series database. It cyclically records characteristic data of all circuits at 1-millisecond intervals. When serious faults such as short circuits, overloads, or equipment burnout occur, the digital twin can automatically extract all raw data from the 5 minutes prior to the fault, combine it with pre-defined anomaly detection rules for deep learning analysis, reconstruct the fault evolution process, and provide a scientific basis for accident liability determination and equipment improvement.

[0052] At the system level, the central processing server has added an energy dispatch optimization module and interconnected with the data interfaces of the photovoltaic power generation system and energy storage battery system in the plant area. When performing step 2, the mobile sampling system, in addition to collecting the power consumption characteristics of the equipment, also simultaneously monitors the voltage and frequency fluctuation characteristics at the sampling points to assess the equipment's sensitivity to microgrid frequency shifts.

[0053] During the intelligent planning process, when generating power supply circuit schemes, the system automatically identifies equipment groups suitable for connection to the DC distribution network or with the potential to participate in demand response. For these devices, the system specifically marks the installation location of the bidirectional metering meters and the wiring requirements of the communication interfaces in the digital work order.

[0054] When calculating cable specifications, the system incorporates a life-cycle cost analysis logic. In addition to considering the initial purchase cost, the system also combines the equipment load curve measured in step 2 and the expected electricity price to calculate the energy loss cost of different cable specifications over the next 10 years. The system ultimately recommends the optimal specification that minimizes the sum of initial investment and long-term losses. This is often a level higher than specifications that only meet safety requirements, but is more economical in long-term operation.

[0055] Once delivered, the digital twin became a core component of the virtual power plant within the plant area. It utilizes the "complete set of legal steady-state characteristic states" in pre-defined anomaly detection rules to identify idle capacity within the plant area in real time. When the power grid issues a peak-shaving command, the digital twin, based on the discrete characteristic fingerprints of each device, accurately calculates the available power load that can be released by adjusting the operating speed of the equipment without affecting production quality, and automatically generates dispatch instructions to be sent to the on-site execution agencies.

[0056] Furthermore, the digital twin in this embodiment also possesses self-evolution capabilities. As operating time increases, the system continuously collects real-world operational data and uses machine learning algorithms to automatically correct the anomaly detection threshold and line loss calibration coefficient generated in step 3. This continuous iterative optimization ensures that the power supply system remains in optimal operating condition throughout its entire lifecycle.

[0057] In summary, through the technical means demonstrated in the above embodiments, this invention transforms complex power supply engineering in plant areas from a traditional model relying on manual experience to a data-driven, algorithm-based digital model. The deep integration of the system's hardware and software logic not only improves the accuracy of pre-construction planning and ensures the quality of the construction process, but also provides irreplaceable technical support for the long-term stable operation and intelligent maintenance of the plant area through the delivery of digital twins.

[0058] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A power supply method for factory areas, characterized in that, Includes the following steps: Step 1, Equipment and Sampling Point Association Steps: Read the factory layout drawings through the central processing server, analyze and extract the building axes, equipment installation reference points and cable tray paths, and establish the factory coordinate system; use identification technology to bind the physical equipment on site with the equipment identification on the drawings, obtain the three-dimensional coordinates of the equipment through the spatial positioning system, and establish a file containing identification code, equipment type, location and rated power for each piece of equipment to be powered in the initial digital twin base map; Step 2, Discrete Feature Sampling Step: Based on the three-dimensional coordinates of the equipment, the optimal temporary sampling points are planned using a cluster center iterative strategy; the full-range excitation protocol is executed through a mobile sampling system to control the equipment to traverse standby, half-load, and full-load states; the real-time fluctuation rate of power consumption characteristics is monitored, and after steady state, power consumption feature vectors such as active power and reactive power are collected to construct a set of discrete feature states of the equipment; Step 3, Intelligent Planning Steps: Define a similarity function for comprehensive electrical feature similarity and spatial proximity, and use a clustering algorithm to group loops; extract peak electrical parameters to specify distributor specifications, and combine maximum steady-state current, power supply distance and measured data to calibrate line loss and determine cable specifications; combine discrete feature states of equipment to form a complete set of legal steady-state features and pre-form anomaly detection rules; Step 4, Digital Construction and Delivery: Generate digital work orders based on the plan, compare deviations after construction is completed, and form a digital twin of the power supply system that carries discrete feature fingerprints and anomaly detection rules.

2. The power supply method for a factory area according to claim 1, characterized in that, The identification binding process in step one includes: obtaining the unique identification code of the device using radio frequency identification or QR code scanning, and mapping and associating it with the virtual device object; the spatial positioning system combines laser scanning point cloud data with the drawing for fitting and calibration to ensure that the position deviation is within a preset range, and constructing an initial digital twin base map with spatial topological relationships.

3. The power supply method for a factory area according to claim 1, characterized in that, The full-range excitation and steady-state determination in step two include: the mobile sampling system sends a command to switch the typical working range of the equipment, monitors the power consumption characteristics through a high-precision current transformer and calculates the rolling standard deviation as the volatility; when the volatility is lower than the first preset threshold and the duration reaches the target, it is determined that the system has entered a steady state and triggers the collection of feature snapshots. The power consumption feature vector also includes the phase angle and amplitude of specific order current harmonics.

4. The power supply method for a factory area according to claim 1, characterized in that, The optimal temporary sampling point planning in step two includes: dynamically clustering based on the three-dimensional coordinates of the device, minimizing the sum of squares of the lengths of temporary cables within the cluster, solving for the optimal sampling points using the gradient descent algorithm, and outputting a set of sampling points to reduce the impact of temporary cable impedance on sampling accuracy.

5. The power supply method for a factory area according to claim 1, characterized in that, The circuit grouping in step three includes: assigning weights to electrical characteristic similarity and spatial proximity, using agglomerative hierarchical clustering algorithm to group devices with similar characteristics and close locations into the same circuit; and combining harmonic phase characteristics to group devices with opposite harmonic phases to reduce interference by utilizing the harmonic self-cancellation effect.

6. The power supply method for a factory area according to claim 5, characterized in that, The electrical feature similarity calculation includes: finding the optimal matching feature pair of the discrete feature state sets of two devices, calculating the weighted Euclidean distance mean as the optimal matching distance; quantifying the differences in the electrical characteristics of the devices, and classifying them into the same circuit if the optimal matching distance is less than a preset threshold.

7. The power supply method for a factory area according to claim 1, characterized in that, The specification of the distributor in step three includes: extracting the maximum instantaneous current and starting power of a subset of the same distributor equipment, ensuring that the rated current of the distributor covers the sum of the maximum demand and combining the power consumption simultaneity rate and dynamic redundancy coefficient, with the redundancy coefficient adjusted according to the power consumption simultaneity rate range.

8. The power supply method for a factory area according to claim 1, characterized in that, The cable specification calculation in the above steps includes: calculating the theoretical voltage drop based on the maximum steady-state current and power supply distance, deriving the actual line loss coefficient and calibrating the model based on measured data; and correcting the equivalent resistance based on ambient temperature and cable temperature rise, and selecting a cable cross-sectional area that meets the requirements for current carrying capacity and voltage drop.

9. A power supply system for a factory area, characterized in that, include: The central processing server executes logic such as data parsing and coordinate transformation, and deploys digital twin base map management modules, electrical feature databases, etc., to establish the plant area coordinate system and initial digital twin base map; The mobile sampling system, including an autonomous mobile chassis and a power quality analyzer, executes a discrete feature sampling and excitation protocol to collect and upload power consumption feature vectors. On-site sensing terminals, including identification tags and spatial positioning terminals, establish a mapping between physical and digital spaces; The digital construction guidance terminal uses an augmented reality industrial tablet computer to achieve visual guidance of construction. The intelligent planning engine generates loop grouping and cable specification schemes, and the anomaly detection rule generator constructs a complete set of legal steady-state features and embeds them into the digital twin. The digital twin monitors loop parameters in real time, and triggers alarms and highlights them when anomalies occur.

Images