Energy-saving optimization method and device for intelligent city lighting and energy management

Abstract

The invention provides an energy-saving optimization method and device for intelligent city lighting and energy management. The method comprises the following steps: acquiring multi-source fusion data acquired by a multi-mode sensor configured at a lighting node in real time; carrying out coupling analysis on junction temperature fluctuation characteristics of the lighting equipment and a harmonic spectrum of a power supply system according to the multi-source fusion data, and determining dynamic characteristic parameters of the lighting equipment; constructing a multi-objective optimization function according to the dynamic characteristic parameters; and processing the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy. Through implementation of the scheme of the invention, the finally output multi-domain cooperative control strategy can dynamically adjust illumination output and power supply deployment in real time according to the sensing data, and the overall energy consumption of the illumination system is effectively reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban lighting, and in particular to an energy-saving optimization method and device for intelligent urban lighting and energy management. BACKGROUND

[0002] With the acceleration of urbanization and the deepening of the construction of "smart city", the urban road lighting system is gradually evolving towards intelligence, low carbon and multi-source perception. Especially in large urban transportation hubs, high-density residential areas and key security areas, the lighting system not only undertakes basic lighting tasks, but also needs to respond to environmental changes, traffic density fluctuations and power load dynamics to form an energy efficiency management mechanism highly coupled with the city's running state. For example, in a large traffic interchange area, the lighting system needs to dynamically adjust the light intensity according to the night traffic density and air quality conditions, and respond to multiple factors such as local photovoltaic output, power harmonic interference and equipment aging, to achieve safe, energy-saving and stable lighting control.

[0003] However, the existing urban lighting system is mostly based on static control rules, lacks deep perception and comprehensive analysis capability of multi-source data such as lighting node device state and power quality, and is difficult to realize dynamic adaptation between light efficiency performance and power supply conditions. Especially in the case of device aging and obvious load fluctuation, the traditional control strategy cannot accurately evaluate the actual output efficiency of the light source and its energy consumption performance, resulting in low lighting energy efficiency, lagging light adjustment response, and even causing local uneven illumination and energy waste. SUMMARY

[0004] The present application provides an energy-saving optimization method and device for intelligent urban lighting and energy management, which solves the problem of low lighting energy efficiency caused by the inability to accurately evaluate the actual output efficiency of the light source and its energy consumption performance in related technologies.

[0005] The first aspect of the present application provides an energy-saving optimization method for intelligent urban lighting and energy management, which comprises: acquiring multi-source fusion data collected by a multi-modal sensor configured in a lighting node in real time; coupling and analyzing the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system according to the multi-source fusion data to determine the dynamic characteristic parameters of the lighting device; constructing a multi-objective optimization function according to the dynamic characteristic parameters; processing the multi-objective optimization function to generate an optimization control strategy in multi-domain coordination.

[0006] Optionally, in the first implementation manner of the first aspect of the present application, the step of acquiring multi-source fusion data collected by a multi-modal sensor configured in a lighting node in real time comprises: Collecting a road surface illuminance distribution parameter by an arrayed photosensor configured at a lighting node, and synchronously acquiring a power waveform characteristic parameter; According to geographical position information of the lighting node, the road surface illuminance distribution parameter is reorganized in a grid, and the power waveform characteristic parameter is subjected to Fourier transform to generate an associated data set of a time-frequency characteristic matrix and a spatial illuminance matrix; The associated data set is cleaned of abnormal data by a dynamic threshold model of device historical operation data to generate a spatio-temporal fusion data set; The spatio-temporal fusion data set is input into a federated learning framework, missing data is interpolated and compensated by a parameter exchange protocol between edge computing nodes to generate a multi-source fusion data set of all the lighting nodes.

[0007] Optionally, in the second implementation manner of the first aspect of the application, the step of coupling analysis of a junction temperature fluctuation characteristic of the lighting device and a harmonic spectrum of the power supply system according to the multi-source fusion data to determine a dynamic characteristic parameter of the lighting device comprises: Multi-scale decomposition is performed on the junction temperature fluctuation characteristic in the multi-source fusion data to extract a junction temperature modal feature set; Energy distribution analysis is performed on the harmonic spectrum of the power supply system to generate a harmonic energy density spectrum feature; Tensor space mapping is performed on a time-domain evolution law of the junction temperature modal feature set and a frequency-domain distribution characteristic of the harmonic energy density spectrum feature to generate a cross-domain coupling feature matrix; Transient response amplitude and trend change rate in the cross-domain coupling feature matrix are input into a nonlinear degradation mapping function to generate a dynamic characteristic parameter set.

[0008] Optionally, in the third implementation manner of the first aspect of the application, the dynamic characteristic parameter comprises a luminous efficacy decay rate, a voltage sensitivity factor, and a thermal resistance dynamic parameter, and the step of constructing a multi-objective optimization function according to the dynamic characteristic parameter comprises: An interval clustering is performed on the luminous efficacy decay rate, the voltage sensitivity factor, and the thermal resistance change curve to construct a mapping matrix of device performance degradation; According to change rates and stable boundary values of each sub-section in the mapping matrix, a joint state of a current device is subjected to state mapping to generate a parameter attribution vector converted from a state label to a target function variable; The parameter attribution vector is time-series synchronously embedded with an illuminance demand matrix, power load prediction data, and a historical abnormal response label to generate a multi-objective coupling structure; Each target function sub-element in the multi-objective coupling structure is subjected to weighted combination to generate a target priority assignment vector. According to the fluctuation sensitivity and frequency-related factors of each target dimension in the target priority assignment vector, the energy consumption term, response time term and illumination deviation term in the objective function are segmented and reconstructed to construct a multi-objective optimization function.

[0009] Optionally, in a fourth implementation of the first aspect of the present application, the step of processing the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy includes: Decomposing the multi-objective optimization function into a device-level dynamic optimization sub-function and a system-level steady-state optimization sub-function; Performing dynamic parameter space compression on the device-level dynamic optimization subfunction, iteratively shrinking the feasible domain of the device control parameters by dynamically adjusting the compression rate of the parameter search space, and generating a device-level dynamic optimization solution set; Performing distributed coupling constraint processing on the system-level steady-state optimization subfunction to generate a system-level steady-state optimization solution set; Multi-objective collaborative decision-making is performed on the device-level dynamic optimization solution set and the system-level steady-state optimization solution set, and the optimization conflicting objectives are prioritized based on the non-dominated solution screening mechanism to generate a multi-domain collaborative optimization control strategy including pulse width modulation instructions and power supply phase compensation parameters.

[0010] Optionally, in a fifth implementation of the first aspect of the present application, after the step of processing the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy, the method further includes: Inputting the multi-domain collaborative optimization control strategy into the digital twin model; Performing spatiotemporal synchronization mapping on the pulse width modulation instruction and the power supply phase compensation parameter through a virtual execution interface to generate a virtual control instruction set; Dynamically modifying parameters of the virtual control instruction set based on multi-source fusion data collected in real time; The modified instruction set is subjected to co-simulation of illumination mutation and load fluctuation to generate a prediction response matrix. The power supply phase compensation parameters are iteratively weighted according to the predicted response matrix to generate a feedback correction control strategy.

[0011] Optionally, in a sixth implementation of the first aspect of the present application, the method further includes: Constructing equipment health evaluation indicators based on the multi-domain collaborative optimization control strategy; Determining a maintenance priority sequence for each lighting node by performing spectral decomposition on the correlation between the equipment health evaluation index and the historical operating parameters; Performing distributed weighted mapping on the maintenance priority sequence to generate a predictive maintenance plan including equipment status warning indicators and maintenance time windows; The configuration parameters of the multimodal sensor are adaptively adjusted according to the predictive maintenance scheme to determine a dynamic allocation strategy for the sampling frequency and monitoring accuracy of the sensor.

[0012] A second aspect of the present application provides an energy-saving optimization device for smart city lighting and energy management, the energy-saving optimization device for smart city lighting and energy management comprising: An acquisition module is used to acquire multi-source fusion data collected in real time by multi-modal sensors configured at lighting nodes; an analysis module, configured to perform a coupling analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data, and determine the dynamic characteristic parameters of the lighting device; A construction module, configured to construct a multi-objective optimization function according to the dynamic characteristic parameters; A generation module is used to process the multi-objective optimization function and generate a multi-domain collaborative optimization control strategy.

[0013] A third aspect of an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the processor is used to execute a computer program stored on the memory. When the processor executes the computer program, it implements each step of the energy-saving optimization method for smart city lighting and energy management provided in the first aspect of the embodiment of the present application.

[0014] The fourth aspect of the embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the energy-saving optimization method for smart city lighting and energy management provided in the first aspect of the embodiment of the present application are implemented.

[0015] In summary, according to the energy-saving optimization method and device for smart city lighting and energy management provided by the present application, multi-source fusion data collected in real time by multi-modal sensors configured at lighting nodes is obtained; coupling analysis of the junction temperature fluctuation characteristics of the lighting equipment and the harmonic spectrum of the power supply system is performed based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting equipment; a multi-objective optimization function is constructed based on the dynamic characteristic parameters; the multi-objective optimization function is processed to generate a multi-domain collaborative optimization control strategy. Through the implementation of the present application, the multi-domain collaborative control strategy ultimately outputted can dynamically adjust the lighting output and power supply allocation in real time based on the perception data, effectively reducing the overall energy consumption of the lighting system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of an energy-saving optimization method for smart city lighting and energy management provided in an embodiment of the present application; Figure 2A program module schematic diagram of the energy-saving optimization device for intelligent city lighting and energy management provided by the embodiment of the present application is shown in the following figure. Figure 3 A structure schematic diagram of the electronic device provided by the embodiment of the present application is shown in the following figure. DETAILED DESCRIPTION

[0017] In order to make the purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0018] In order to solve the problem that the actual output efficiency of the light source and its energy consumption performance cannot be accurately evaluated in the related art, resulting in low lighting energy efficiency, the present application provides an energy-saving optimization method for intelligent city lighting and energy management, as shown in the following figure. Figure 1 A flowchart of the energy-saving optimization method for intelligent city lighting and energy management provided by the embodiment is shown in the following figure. The energy-saving optimization method for intelligent city lighting and energy management includes the following steps: Step 110, acquiring multi-source fusion data collected by the multi-modal sensor configured in the lighting node in real time.

[0019] Specifically, by configuring the multi-modal sensor in the lighting node, high-frequency and wide-coverage data collection of the urban road operating environment is realized. The sensor system usually includes an illumination intensity sensor, a temperature and humidity module, a PM2.5 detector, a voltage / current transformer and a vehicle flow monitoring unit. The above sensors constitute a multi-source perception network, and the collected data is uploaded to the edge computing node or the cloud platform through the LoRaWAN or 5G communication link. In order to improve the data quality, a sliding window mechanism and an anomaly detection algorithm are deployed at the edge to perform real-time cleaning and standardization processing on the original data stream, and then a structured spatio-temporal fusion data set is generated to ensure that the lighting system has comprehensive and real-time perception ability for external environment and internal equipment state.

[0020] In an optional implementation of the embodiment, the step of acquiring the multi-source fusion data collected by the multi-modal sensor arranged in the lighting node in real time comprises: collecting a road surface illuminance distribution parameter by an array photoelectric sensor arranged in the lighting node, and synchronously acquiring a power waveform characteristic parameter; performing grid reorganization on the road surface illuminance distribution parameter according to geographical position information of the lighting node, and performing Fourier transform on the power waveform characteristic parameter to generate an associated data set of a time-frequency characteristic matrix and a spatial illuminance matrix; performing cleaning on abnormal data of the associated data set by a dynamic threshold model of device historical operation data to generate a space-time fusion data set; and inputting the space-time fusion data set into a federated learning framework, performing interpolation compensation on missing data by a parameter exchange protocol between edge computing nodes to generate a multi-source fusion data set of all the lighting nodes.

[0021] Specifically, by installing array photoelectric sensors at lighting nodes, real-time collection of road illumination distribution is achieved. The sensor is composed of multiple photosensors and can capture the illumination intensity in different directions. The data acquisition system works synchronously with the voltage / current transformer, which can simultaneously record the time domain characteristics of the power waveform. The voltage / current transformer is a device that converts high voltage or high current signals into low amplitude signals and is used to monitor the status of the power system, thereby providing the original data basis for subsequent data processing. Secondly, based on the positioning information of the lighting nodes, the collected road illumination data is grid-reorganized to convert the continuous spatial distribution data into a grid. The data is divided into several grid cells based on geographic regions, forming a spatial illumination matrix that reflects the illumination distribution of different road areas. Simultaneously, a sliding window Fourier transform is applied to the power waveform data, converting the signal from the time domain to the frequency domain and capturing temporal variations. This generates a time-frequency feature matrix. The spatial illumination matrix generated at the same timestamp is paired with the corresponding time-frequency feature matrix to construct a three-dimensional tensor structure, thereby linking data of different dimensions. The third dimension represents different perceptual modalities (illuminance, voltage, and frequency). For example, this transformation can identify anomalous spectral components in the power waveform caused by harmonic distortion. A dynamic threshold model constructed using historical equipment operating data is used to remove anomalous data from the aforementioned linked dataset. This model automatically sets judgment criteria based on historical statistical characteristics and can clean up anomalous signals when data deviates from the expected range. For example, a sudden increase or decrease in the illumination value of a grid cell is considered inconsistent with the actual operating state. This generates a spatiotemporal fusion dataset, ensuring the accuracy and continuity of data information. The standardized data set is input into the federated learning framework, which implements collaborative data processing in a distributed environment. Through a parameter exchange protocol between edge computing nodes, interpolation is used to compensate for missing data, allowing data from each lighting node to be shared and integrated, thus forming a globally unified multi-source fusion dataset. This dataset provides sufficient data support for subsequent intelligent control algorithms, effectively improving the dynamic response capabilities of urban lighting systems to the environment and equipment status.

[0022] It should be noted that federated learning is a machine learning paradigm that ensures data remains local, only model parameters are transferred. It is particularly well-suited for deployment across the large number of edge computing nodes found in lighting systems. In this system, each edge node maintains a local, standardized subset of data and uses it to train a local sub-model, such as a lightweight neural network for predicting junction temperature trends or illuminance response delays. During training, each node does not upload raw data. Instead, it sends the model gradients or weight updates generated by each training round to the cloud or a central coordination server via a secure channel. After receiving the model parameters from multiple nodes, the server uses a weighted aggregation algorithm (such as FedAvg) to fuse all parameters into a unified global model. This fused model is then distributed to each node for the next round of training. If some nodes have missing data, such as due to temporary sensor offline or communication interruption, the system uses model parameters from other nodes to perform gradient interpolation on their local models, thereby indirectly recovering the missing data while maintaining model accuracy. Ultimately, the federated learning framework enables the sharing of model capabilities and data semantics across multiple nodes without leaking local private data, generating a multi-source fused dataset with broad adaptability and strong contextual relevance.

[0023] Step 120: Perform coupling analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting device.

[0024] Specifically, based on the standardized multi-source data obtained, joint modeling and coupled analysis are performed on the junction temperature fluctuation characteristics of lighting equipment and the harmonic spectrum of the power supply system. Through time series decomposition and modal identification of junction temperature data, the temporal variation patterns reflecting thermal characteristics are extracted. Simultaneously, Fourier transform and wavelet packet analysis techniques are used to characterize the harmonic energy density in the power waveform in the frequency domain, thereby identifying potential power supply anomalies. By tensor mapping the thermal modal characteristics with the spectral characteristics, a cross-domain coupling matrix is ​​constructed. This is further input into a nonlinear degradation mapping model to derive a set of dynamic characteristic parameters reflecting the light decay rate, voltage sensitivity, and thermal resistance changes, laying the data foundation for subsequent modeling.

[0025] In an optional implementation of this embodiment, coupling analysis is performed on the junction temperature fluctuation characteristics of the lighting equipment and the harmonic spectrum of the power supply system based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting equipment, including: performing multi-scale decomposition on the junction temperature fluctuation characteristics in the multi-source fusion data to extract a junction temperature modal feature set; performing energy distribution analysis on the harmonic spectrum of the power supply system to generate harmonic energy density spectrum characteristics; generating a cross-domain coupling feature matrix by performing tensor space mapping on the time domain evolution law of the junction temperature modal feature set and the frequency domain distribution characteristics of the harmonic energy density spectrum characteristics; inputting the transient response amplitude and trend change rate in the cross-domain coupling feature matrix into a nonlinear degradation mapping function to generate a dynamic characteristic parameter set.

[0026] Specifically, the junction temperature of lighting devices, i.e. the temperature of the internal junction of semiconductor light-emitting diodes (LEDs), will exhibit dynamic characteristics that fluctuate over time during long-term operation, reflecting the coupling relationship between the changes in the thermal environment and the thermal degradation behavior of the device. In order to more accurately describe the internal rules of the evolution of junction temperature over time, it is necessary to perform multi-scale decomposition on the collected junction temperature time series signals. Multi-scale decomposition is a time-domain analysis method that processes original non-stationary signals in different time scales, often achieved through methods such as wavelet packet decomposition, which can extract modal information reflecting the thermal inertia, sudden temperature rise, and cooling response of the device at different resolutions. Taking the junction temperature changes of lighting devices during morning and evening rush hours on urban roads as an example, the influence of sunlight and environmental temperature on the thermal accumulation trend is revealed at a long scale, while the junction temperature jump response within a short time after the streetlights are turned on is revealed at a short scale. This multi-level extraction method can better preserve the intrinsic characteristics of the signal compared to simple statistical averaging or extreme value analysis, thereby improving the accuracy of the thermal state description. In the process of analyzing the impact of power quality on the operating state of lighting devices, it is necessary to analyze the harmonics that may exist in the power signal in detail. Harmonics refer to frequency components that are integer multiples of the fundamental frequency, which are usually caused by nonlinear loads and can have a negative impact on grid stability and device performance. To accurately identify the degree of harmonic influence, the voltage or current signal is processed through fast Fourier transform (FFT) to obtain its frequency spectrum structure, and by calculating the energy density distribution of each frequency component, a harmonic energy density spectrum feature can be generated. This spectrum feature reflects the energy concentration at different frequency bands, for example, when an abnormal energy peak appears in the third harmonic frequency band, it indicates that there may be a high-frequency disturbance source causing electrical waveform distortion. To further reveal the interaction mechanism between thermal characteristics and power disturbances, the junction temperature modal features extracted earlier are fused with the harmonic energy density spectrum features through tensor space mapping. Tensor space mapping is a method of projecting multiple heterogeneous dimensional data into a unified high-dimensional structure, which is used here to construct a cross-time and frequency domain data fusion expression model. By constructing a three-dimensional tensor, one dimension represents time evolution, another dimension represents frequency response, and the third dimension represents the spatial index of the perception node, a multi-dimensional cross-expression of multiple source state parameters of the lighting system can be achieved. For example, at a lighting node on a certain road section, when the modal feature of rapid short-term rise in junction temperature and the concentration of harmonic energy in a specific frequency band occur simultaneously, it will appear as a local high-intensity block structure in the tensor mapping, which helps to identify the thermal load abnormalities induced by power grid disturbances. The transient response amplitude and trend change rate in the cross-domain coupled feature matrix constructed are input into a nonlinear degradation mapping function to describe the performance degradation process of the lighting device under dynamic interference conditions. The mapping function is established based on a nonlinear regression model or a neural network framework, with the key coupling indicators in the tensor matrix as input variables and a set of dynamic characteristic parameters representing the operating state of the device, such as light efficiency decay rate, voltage sensitivity factor, and thermal resistance dynamic parameters, as output variables.The light efficiency decay rate reflects the light output change trend under unit power consumption, the voltage sensitivity factor depicts the steady-state response degree of the device to voltage fluctuation, and the thermal resistance dynamic parameter is used to quantify the trend of the heat conduction efficiency changing with the use time. Compared with relying only on the static characteristic curve provided by the manufacturer, this method can realize dynamic modeling of the device state under specific operating scenarios, thereby providing more accurate and responsive technical support in actual applications.

[0027] Step 130, constructing a multi-objective optimization function according to the dynamic characteristic parameters.

[0028] Specifically, a multi-objective optimization function for a city lighting system is constructed according to the above dynamic characteristic parameters. The function comprehensively considers key indicators such as lighting energy consumption, power grid stability, and user comfort, and embeds dynamic constraint conditions related to device operating state and environmental conditions on this basis. The function expression is established in a weighted linear combination or mixed integer programming manner, and the priority management between objectives is realized by adjusting the weight coefficient of the objective function.

[0029] In an optional embodiment of the present embodiment, the step of constructing a multi-objective optimization function according to the dynamic characteristic parameters includes: constructing a mapping matrix of device performance degradation by interval clustering of the light efficiency decay rate, the voltage sensitivity factor, and the thermal resistance change curve; performing state mapping of the joint state of the current device according to the change rate and stable boundary value of each sub-section in the mapping matrix, generating a parameter attribution vector that converts the state label to the objective function variable; generating a multi-objective coupling structure by time sequence synchronization embedding of the parameter attribution vector, the illumination demand matrix, the power load prediction data, and the historical abnormal response label; performing weighted combination of each objective function sub-element in the multi-objective coupling structure to generate an objective priority assignment vector; and performing segmented reconstruction of the energy consumption item, the response time item, and the illumination deviation item in the objective function according to the fluctuation sensitivity and frequency related factors of each objective dimension in the objective priority assignment vector, to construct a multi-objective optimization function.

[0030] Specifically, in the process of constructing the multi-objective optimization function, the device state data with discontinuous variation characteristics can be converted into feature sections with identifiable boundaries by interval clustering of the light efficiency decay rate, voltage sensitivity factor and thermal resistance change curve. The time series of the above three types of dynamic parameters are segmented and clustered according to the trend of change, so that each sub-section has a consistent evolution slope and disturbance amplitude, which can identify the performance mutation points and stable intervals in modeling. For example, for the obvious decline segment of light efficiency after 3000 hours and 6000 hours, the corresponding high degradation zone and slow change zone can be constructed to reflect the nonlinear degradation process of device performance. The performance degradation mapping matrix established in this way has multiple variable input dimensions and segmented clustering indexes, and can present the state transition characteristics of the device between different service stages. By analyzing the change rate and stable boundary value of each sub-section in the mapping matrix, the position of the current performance section of each device in the dynamic characteristic space can be obtained, and then the joint state of the device is mapped. State mapping refers to projecting the three types of features of light decay, voltage disturbance and thermal resistance growth into a discrete state space. Each state node is labeled with a specific state label, such as "low thermal resistance-high electrical sensitivity-medium light decay" or "high thermal resistance-low electrical sensitivity-high light decay", and a mapping relationship from state label to optimization variable is established based on this. In the actual system, if a device is in a high thermal resistance and strong light decay combined state, it can be mapped to a high load constraint state node, and its optimization weight can be attributed to the energy consumption and response time dimensions. Subsequently, the parameter attribution vector needs to be time-synchronously embedded with the illumination demand matrix, power load prediction data and historical abnormal response label to construct a multi-objective coupling structure. The illumination demand matrix represents the lighting level demand at different time and space nodes, the power load prediction data reflects the fluctuation trend of the load in the future interval, and the historical abnormal response label records the deviation pattern caused by unstable response of the device in the past. By aligning the three types of data through time stamp and constructing tensor nested structure, the state of each device can be accurately coupled with its power supply background, illumination demand and risk history. For example, during the peak traffic period, if the device state label points to the high light decay section, and the load prediction shows a rising trend in the future load, and the abnormal label appears multiple light transients, then the structure formed by the device will strengthen the representation ratio of the illumination dimension and the stability dimension in the objective function. Based on the statistical characteristics and weight distribution of each objective function sub-element in the coupled structure, a weighted combination can be performed to generate a target priority assignment vector. This vector contains the relative importance expression of each target dimension under the current system state, quantifying the mutual exclusion and synergy degree of energy minimization, response time control and illumination uniformity. In the process of weighted combination, not only the assignment of the parameter attribution vector is referred to, but also the illumination boundary limit and load fluctuation confidence are combined, so as to allocate a dynamic weight based on state sensitivity to each sub-objective.Taking lighting nodes in accident-prone road sections as an example, under low-voltage fluctuations at night, the weight of the response time dimension will be significantly increased due to the frequent occurrence of abnormal labels, while the energy consumption dimension will be weakened due to the slower slope of light efficiency decay. This approach can dynamically adjust the target priority sequence and guide the optimization solution to diverge towards higher-risk dimensions in urban lighting networks with changing constraints. Finally, based on the fluctuation sensitivity and frequency-related factors corresponding to each target dimension in the target priority assignment vector, the energy consumption, response time, and illumination deviation terms in the objective function can be segmented and reconstructed. The fluctuation sensitivity quantifies the amplitude of the nonlinear response of each target sub-element to input perturbations, while the frequency factor describes the frequency of its occurrence as the dominant target in the historical operation sequence. By introducing a non-uniform weighting function and a threshold filtering mechanism, the original target term is segmented and the function is transformed. For example, a segmented penalty function is introduced for the illumination deviation term, resulting in a penalty growth function in high-traffic areas and a smooth transition to a linear term in green belt areas. This generates an optimization function expression with temporal and spatial differentiation.

[0031] Step 140: Process the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy.

[0032] Specifically, a hierarchical optimization engine processes the multi-objective optimization function, first dividing the optimization problem into a device-level dynamic optimization sub-function and a system-level steady-state optimization sub-function, thereby achieving hierarchical decoupling and efficient division of labor for computational tasks. The device-level sub-function employs a dynamic parameter space compression method, adjusting the compression ratio of the parameter search space based on real-time operating conditions to gradually converge the feasible domain of control parameters and obtain a more responsive dynamic optimization solution set. Simultaneously, the system-level sub-function incorporates a distributed coupling constraint processing strategy. Based on the energy coupling relationships between different nodes and network operation requirements, a steady-state optimization solution is solved in a distributed architecture to obtain a system-level steady-state optimization solution set. Subsequently, a multi-objective collaborative decision-making mechanism is used to integrate the device-level and system-level solutions. Conflicting objective functions are prioritized based on the non-dominated solution screening principle. Ultimately, a multi-domain collaborative optimization control strategy, including pulse width modulation instructions and power supply phase compensation parameters, is output, achieving optimal energy-efficient scheduling of the urban lighting system under different operating conditions.

[0033] In an optional implementation of this embodiment, the multi-objective optimization function is processed to generate a multi-domain collaborative optimization control strategy, including: decomposing the multi-objective optimization function into a device-level dynamic optimization sub-function and a system-level steady-state optimization sub-function; performing dynamic parameter space compression on the device-level dynamic optimization sub-function, iteratively shrinking the feasible domain of the device control parameters by dynamically adjusting the compression rate of the parameter search space, and generating a device-level dynamic optimization solution set; performing distributed coupling constraint processing on the system-level steady-state optimization sub-function to generate a system-level steady-state optimization solution set; performing multi-objective collaborative decision-making on the device-level dynamic optimization solution set and the system-level steady-state optimization solution set, prioritizing the optimization conflicting objectives based on the non-dominated solution screening mechanism, and generating a multi-domain collaborative optimization control strategy including pulse width modulation instructions and power supply phase compensation parameters.

[0034] Specifically, by decomposing the multi-objective optimization function, the overall problem is divided into dynamic sub-functions targeting individual device responses and system-level sub-functions reflecting the steady-state behavior of the entire network. This achieves a hierarchical solution, from local dynamic control to global steady-state equilibrium. This decomposition approach enables the separate solution of complex optimization problems at different scales. For example, in an intelligent lighting system, the dimming response of a single streetlight and the load balance of the entire regional power grid can be modeled separately, facilitating the development of targeted control strategies. Dynamic parameter space compression technology is then introduced to the device-level dynamic sub-function. By monitoring the current solution status of device control parameters, the compression ratio of the parameter search space (the ratio or intensity of the reduction in the parameter range in the optimization algorithm) is adaptively adjusted. This iteratively shrinks the parameter feasible region, thereby continuously focusing the search on the optimal solution region. The device-level dynamic optimization solution set refers to the set of optimal solutions obtained by dynamically optimizing the control parameters of each lighting device under varying operating conditions (such as ambient light intensity, temperature, and power grid fluctuations). The device-level dynamic optimization solution set includes, but is not limited to, pulse width modulation control instructions, drive current adjustment parameters, and predicted results of device state variables. Subsequently, a distributed coupled constraint processing strategy is employed to solve the system-level steady-state subfunction. Through a constraint transfer algorithm based on physical network topology and energy consumption coordination, and leveraging inter-regional data exchange and collaborative solution mechanisms, global constraints are jointly processed. This approach ensures that local decisions regarding grid load balancing and energy allocation remain consistent and meet global optimality requirements. For example, in a multi-region power supply system, inter-node constraint coupling can effectively balance power fluctuations. The system-level steady-state optimization solution set includes, but is not limited to, power phase compensation parameters, regional lighting allocation strategies, and grid coupling control parameters. After obtaining the solution sets obtained at the device level and the system level respectively, the two parts of the solution sets are integrated through a multi-objective collaborative decision-making mechanism. This decision-making process is based on the non-dominated solution screening mechanism. This method adopts the Pareto optimal theory to prioritize conflicting objectives from different optimization scales, thereby achieving an effective combination of pulse width modulation instructions and power supply phase compensation parameters. The combined output is a multi-domain collaborative optimization control strategy, which enables the entire intelligent lighting system to meet the dynamic response requirements of a single device while maintaining a balance between global energy efficiency and grid stability. For example, in the lighting system of a certain area, if the dynamic solution set at the device level recommends high brightness output, but the steady-state solution set at the system level determines that the grid load is close to the threshold, then the collaborative decision-making module will select a balanced solution according to the non-dominated sorting mechanism, such as slightly reducing the brightness and synchronously adjusting the power supply phase, ultimately forming an executable, feedback-enabled, and scalable control strategy.

[0035] It should be noted that the multi-domain collaborative optimization control strategy is essentially a multi-objective nonlinear optimization process that combines a device-level dynamic optimization subset with a system-level steady-state optimization subset. Its essence is to cross-map and prioritize the non-dominated solution sets of the multi-objective function, outputting a set of control parameter combinations that satisfy both local dynamics and global steady-state requirements.

[0036] In an optional implementation of this embodiment, the multi-domain collaborative optimization control strategy is input into the digital twin model, that is, the control parameters and decision instructions obtained through hierarchical optimization are imported into a virtual platform that constructs a digital image of the physical device and its operating environment. The digital twin model refers to a virtual simulation system established using real-time data. By accurately replicating the actual device status and operating conditions, a simulation environment is provided for instruction verification and subsequent adjustment. For example, in an urban lighting system, the digital twin model can accurately reproduce information such as road illumination, device temperature, and grid load. Then, the virtual execution interface performs spatiotemporal synchronous mapping of the pulse width modulation instructions and the power supply phase compensation parameters. By matching the control signals according to the time sequence and the geographic space coordinates, the original instructions are converted into a virtual control instruction set with a synchronous execution effect in the virtual environment, ensuring that the physical instructions can match the actual operating status on the simulation platform. For example, in an intersection area, the spatial synchronous mapping of the instructions ensures that each street lamp receives accurate dimming instructions. Subsequently, the virtual control instruction set is dynamically modified based on the multi-source fusion data collected in real time, and the sensor is used to perform the dynamic parameter correction. The system updates model parameters based on sensor data and fine-tunes control instructions to compensate for deviations from actual conditions. For example, when the outdoor illuminance is detected to be lower than expected, the system automatically enhances dimming commands to ensure a high degree of consistency between the virtual instructions and the measured data, thus ensuring higher accuracy of the control strategy before execution. Furthermore, the revised instruction set is subjected to a co-simulation of sudden illumination changes and load fluctuations. A dynamic simulation algorithm is used to jointly simulate sudden illumination changes and power load fluctuations, generating a predicted response matrix that contains spatiotemporal information about device responses under different abnormal operating conditions. This predicted response matrix helps to identify potential operational anomalies of the lighting system in special scenarios in advance. For example, in extreme weather conditions such as heavy rain, the prediction of illuminance and load responses can provide a guide for optimizing strategy adjustments. Finally, the power supply phase compensation parameters are iteratively weighted based on the predicted response matrix to generate a feedback correction control strategy. Through multiple rounds of iteration, the system can adaptively optimize the phase compensation parameters, making the feedback correction control strategy more consistent with actual operating requirements, thereby achieving stable, efficient, and safe operation of the lighting system in the long term. It should be noted that the weights determine the priority and response amplitude of different control objectives in the final compensation strategy, allowing for a dynamic balance between multi-objective optimizations (such as illumination uniformity, power quality, equipment life, etc.). Specifically, when a certain objective (such as lighting efficiency stability) is more important in the current scenario, its dominant role in the optimization of phase compensation parameters can be enhanced by increasing its corresponding weight. The compensation parameters generated in this way will be more inclined to safeguard that objective, enabling the system to be adaptive under different operating conditions. For example, during peak traffic hours at night, the system may increase the weight of illumination uniformity and adjust the phase compensation accordingly to minimize lighting fluctuations; while during low-load periods, the system may pay more attention to energy efficiency, and the weight adjustment will cause the compensation parameters to be optimized to the minimum loss configuration.

[0037] In an optional implementation of this embodiment, within the framework of a multi-domain collaborative optimization control strategy, a device health assessment metric with state-responsive tracking capabilities is constructed by multi-dimensionally coupling device dimming parameters, power supply strategies, energy storage scheduling behaviors, and operating state characteristics. This metric comprehensively considers the device's lumen depreciation ratio, thermal resistance margin, and output stability under voltage fluctuations, and introduces a dynamic normalization factor to enable comparability between devices with different ratio levels. Lumen depreciation measures the ratio of the device's current output illuminance to the factory standard illuminance. Thermal resistance margin reflects the chip's ability to buffer thermal fluctuations from the rate of rise of junction temperature to ambient temperature, and voltage stability is calculated by monitoring the rate of change of lighting power in the high THD (Total Harmonic Distortion) region of the power supply waveform. This evaluation metric uses a normalized mapping function to project these three parameters onto a unified health score range, enabling a quantitative description of each device's state within a complex operating context. By performing spectral decomposition analysis on the correlation between equipment health assessment indicators and historical operating parameters, periodic degradation modes and mutation anomalies present in long-term operation can be identified. Spectral decomposition involves performing a joint singular value decomposition (SVD) on the health indicator series with historical series of continuous variables such as temperature, current, and illuminance deviation. This process extracts characteristic frequency bands in the principal component spectrum, revealing which historical parameter combinations have the dominant influence on health changes over time. This analysis can reveal typical equipment degradation behaviors in specific seasons, during periods of severe load fluctuations, or near high traffic areas. For example, if a node in a certain area exhibits a significant frequency band of sudden light attenuation during high temperature and high humidity in the summer, this frequency band can be included as the primary degradation mode in the equipment profile. After obtaining the dominant spectral characteristics, vector distance mapping can be used to cluster different lighting nodes. A maintenance priority sequence can then be constructed based on mode severity and mutation frequency, ensuring that devices with high degradation frequency bands receive higher priority for maintenance scheduling when system resources are limited. Subsequently, by subjecting the maintenance priority sequence to distributed weighted mapping, a predictive maintenance plan can be generated that includes equipment status warning indicators and maintenance time windows. Weighted mapping constructs a multidimensional weight vector based on each node's priority, the subsystem load density of the device's location, and historical anomaly response intervals. The strategy mapping function is then distributed across edge computing units. Maintenance windows are generated based on the intersection of the slope growth rate of the historical health fluctuation range and the tangent line of the degradation prediction curve. The greater the slope and the closer the intersection, the earlier the window is assigned. Warning indicators are also weighted and adjusted based on the functional level of the node's area. For example, urban main roads and areas surrounding schools will receive earlier windows under the same health conditions.After constructing the dual axes of maintenance weight and time window, the system integrates these two dimensions to output a joint state-time matrix. Each row corresponds to a specific node, and each column corresponds to its current state feature label, health gradient, warning score, maintenance priority, and recommended time window. This ultimately forms a predictive maintenance plan for system scheduling. For example, consider a lighting node on a road section, numbered LD-3205. Its current luminous efficiency decay rate has reached 45% of its factory default value, its THD harmonic distortion exceeds the standard by 15%, and its thermal resistance curve has shown positive growth for three consecutive cycles. This node is labeled as "medium-high degradation combined state" and has a health score of 0.62. After weighting, its priority is ranked in the top 15% within its region. Further analysis of its historical degradation slope reveals that it is expected to fall below the system trigger threshold of 0.6 within the next 72 hours. Therefore, the system generates a maintenance window of [T+12h, T+48h] for it, indicating that the recommended maintenance period is between 12 hours and two days from the current time. This approach not only establishes a static sequence but also introduces parallel dimensions of time and behavior labels, enhancing the robustness of maintenance decisions under multi-source uncertainty. When resources are limited, the system can perform policy pruning or dynamic priority screening based on this joint matrix, ensuring that key nodes remain within the maintenance intervention zone while edge nodes can be deferred based on window tolerance. Unlike maintenance triggering strategies based on a single scoring function, this matrix-based strategy, based on spatiotemporal synchronization, demonstrates greater adaptability to unevenly distributed and highly coupled lighting system operations.

[0038] Based on the predictive maintenance solution, adaptive adjustment of multimodal sensor configuration parameters dynamically determines the allocation strategy for each sensor's sampling frequency and monitoring accuracy, thereby achieving effective coverage of condition awareness information under conditions of limited sensor resources. This adaptive adjustment mechanism constructs a priority perception model based on the current node's corresponding maintenance window position and the weight of the warning indicator. It also introduces a multimodal redundancy compression algorithm to downsample non-critical parameters. Specifically, for nodes with high warning levels, the light sensor sampling frequency is increased from the original 1Hz to 5Hz, while the PM2.5 and humidity modules maintain their current accuracy to focus on changes in light-emitting unit status. For low-weight nodes with distant maintenance windows, the light module sampling frequency is reduced to 0.2Hz, and model interpolation is used to complement continuous fluctuation trends. Furthermore, for nodes with high voltage fluctuation sensitivity, the frequency domain sampling accuracy of the current transformer is enhanced, for example, from the original 0–1kHz bandwidth to 3kHz to capture abnormal behavior caused by short-term spike disturbances. This strategy does not rely on static perception configuration, but dynamically adjusts the investment of sensing resources based on the evolution of device status. Compared with existing monitoring mechanisms based on unified frequency or manual parameter adjustment, it can enhance the status analysis capabilities of key nodes while controlling communication overhead.

[0039] According to the energy-saving optimization method for smart city lighting and energy management provided by the present application, multi-source fusion data collected in real time by multimodal sensors configured at lighting nodes is obtained; based on the multi-source fusion data, a coupling analysis is performed on the junction temperature fluctuation characteristics of the lighting equipment and the harmonic spectrum of the power supply system to determine the dynamic characteristic parameters of the lighting equipment; a multi-objective optimization function is constructed based on the dynamic characteristic parameters; and the multi-objective optimization function is processed through a hierarchical optimization engine to generate a multi-domain collaborative optimization control strategy. Through the implementation of the present application, the final output multi-domain collaborative control strategy can dynamically adjust the lighting output and power supply allocation in real time based on the perception data, effectively reducing the overall energy consumption of the lighting system.

[0040] Figure 2 The embodiment of the present application provides an energy-saving optimization device for smart city lighting and energy management, which can be used to implement the energy-saving optimization method for smart city lighting and energy management in the aforementioned embodiment. Figure 2 As shown, the energy-saving optimization device for smart city lighting and energy management mainly includes: An acquisition module 10 is used to acquire multi-source fusion data collected in real time by multi-modal sensors configured at lighting nodes; An analysis module 20 is configured to perform a coupled analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting device; A construction module 30 is used to construct a multi-objective optimization function according to the dynamic feature parameters; The generation module 40 is used to process the multi-objective optimization function and generate a multi-domain collaborative optimization control strategy.

[0041] In an optional implementation of this embodiment, the acquisition module is specifically used to: collect road surface illumination distribution parameters through array photoelectric sensors configured at lighting nodes, and synchronously obtain power waveform characteristic parameters; grid-reorganize the road surface illumination distribution parameters according to the geographical location information of the lighting nodes, and perform Fourier transform on the power waveform characteristic parameters to generate an associated data set of a time-frequency feature matrix and a spatial illumination matrix; clean abnormal data of the associated data set through a dynamic threshold model of historical equipment operation data to generate a spatiotemporal fusion data set; input the spatiotemporal fusion data set into a federated learning framework, interpolate and compensate for missing data through a parameter exchange protocol between edge computing nodes, and generate a multi-source fusion data set of all the lighting nodes.

[0042] In an optional implementation of this embodiment, the analysis module is specifically used to: perform multi-scale decomposition on the junction temperature fluctuation characteristics in the multi-source fusion data to extract the junction temperature modal feature set; perform energy distribution analysis on the harmonic spectrum of the power supply system to generate harmonic energy density spectrum characteristics; generate a cross-domain coupling feature matrix by performing tensor space mapping on the time domain evolution law of the junction temperature modal feature set and the frequency domain distribution characteristics of the harmonic energy density spectrum characteristics; input the transient response amplitude and trend change rate in the cross-domain coupling feature matrix into the nonlinear degradation mapping function to generate a dynamic feature parameter set.

[0043] In an optional implementation of this embodiment, the construction module is specifically used to: construct a mapping matrix of device performance degradation by performing interval clustering on the light efficiency decay rate, the voltage sensitivity factor and the thermal resistance change curve; perform state mapping on the joint state of the current device according to the change rate and stable boundary value of each sub-segment in the mapping matrix, and generate a parameter attribution vector that converts the state label to the objective function variable; generate a multi-objective coupling structure by synchronously integrating the parameter attribution vector with the illumination demand matrix, the power load forecast data and the historical abnormal response label; perform weighted combination of each objective function sub-element in the multi-objective coupling structure to generate a target priority assignment vector; and perform segmented reconstruction of the energy consumption item, the response time item and the illumination deviation item in the objective function according to the fluctuation sensitivity and frequency correlation factor of each target dimension in the target priority assignment vector to construct a multi-objective optimization function.

[0044] In an optional implementation of this embodiment, the generation module is specifically used to: decompose the multi-objective optimization function into a device-level dynamic optimization sub-function and a system-level steady-state optimization sub-function; perform dynamic parameter space compression on the device-level dynamic optimization sub-function, iteratively shrink the feasible domain of the device control parameters by dynamically adjusting the compression rate of the parameter search space, and generate a device-level dynamic optimization solution set; perform distributed coupling constraint processing on the system-level steady-state optimization sub-function to generate a system-level steady-state optimization solution set; perform multi-objective collaborative decision-making on the device-level dynamic optimization solution set and the system-level steady-state optimization solution set, prioritize the optimization conflicting objectives based on the non-dominated solution screening mechanism, and generate a multi-domain collaborative optimization control strategy including pulse width modulation instructions and power supply phase compensation parameters.

[0045] In an optional implementation of this embodiment, the generation module is also used to: input the multi-domain collaborative optimization control strategy into the digital twin model; perform spatiotemporal synchronous mapping of the pulse width modulation instructions and the power supply phase compensation parameters through a virtual execution interface to generate a virtual control instruction set; perform dynamic parameter correction on the virtual control instruction set based on multi-source fusion data collected in real time; perform collaborative simulation of illumination mutations and load fluctuations on the corrected instruction set to generate a predicted response matrix; perform iterative weight adjustment on the power supply phase compensation parameters based on the predicted response matrix to generate a feedback correction control strategy.

[0046] In an optional implementation of this embodiment, the generation module is further used to: construct an equipment health assessment index based on the multi-domain collaborative optimization control strategy; determine the maintenance priority sequence of each lighting node by performing spectral decomposition on the correlation between the equipment health assessment index and the historical operating parameters; perform distributed weighted mapping on the maintenance priority sequence to generate a predictive maintenance plan including equipment status warning indicators and maintenance time windows; adaptively adjust the configuration parameters of the multimodal sensor according to the predictive maintenance plan to determine the dynamic allocation strategy of the sensor's sampling frequency and monitoring accuracy.

[0047] According to the energy-saving optimization device for smart city lighting and energy management provided by the present application, multi-source fusion data collected in real time by multimodal sensors configured at lighting nodes is obtained; based on the multi-source fusion data, a coupling analysis is performed on the junction temperature fluctuation characteristics of the lighting equipment and the harmonic spectrum of the power supply system to determine the dynamic characteristic parameters of the lighting equipment; a multi-objective optimization function is constructed based on the dynamic characteristic parameters; and the multi-objective optimization function is processed through a hierarchical optimization engine to generate a multi-domain collaborative optimization control strategy. Through the implementation of the present application, the final output multi-domain collaborative control strategy can dynamically adjust the lighting output and power supply allocation in real time based on the perception data, effectively reducing the overall energy consumption of the lighting system.

[0048] According to the application planFigure 3 An electronic device provided in an embodiment of the present application can be used to implement the energy-saving optimization method for smart city lighting and energy management in the aforementioned embodiment, mainly comprising: Memory 301, processor 302, and computer program 303 stored in memory 301 and executable on processor 302. Memory 301 and processor 302 are connected via communication. When processor 302 executes computer program 303, the energy-saving optimization method for smart city lighting and energy management described in the aforementioned embodiment is implemented. The number of processors can be one or more.

[0049] The memory 301 can be a high-speed random access memory (RAM) memory or a non-volatile memory such as a disk drive. The memory 301 is used to store executable program code. The processor 302 is coupled to the memory 301 .

[0050] Furthermore, the embodiment of the present application also provides a computer-readable storage medium, which can be provided in the electronic device in the above embodiments. The computer-readable storage medium can be the above Figure 3 Memory in the illustrated embodiment.

[0051] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the energy-saving optimization method for smart city lighting and energy management described in the aforementioned embodiments. Furthermore, the computer-readable storage medium may be any medium capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), RAM, a magnetic disk, or an optical disk.

[0052] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0053] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0054] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for energy saving optimization of smart city lighting and energy management, characterized in that: include: Acquire multi-source fusion data collected in real time by multi-modal sensors configured at lighting nodes; performing a coupling analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting device; Constructing a multi-objective optimization function according to the dynamic characteristic parameters; The multi-objective optimization function is processed to generate a multi-domain collaborative optimization control strategy.

2. The energy-saving optimization method for smart city lighting and energy management according to claim 1, characterized in that: The step of obtaining multi-source fusion data collected in real time by a multimodal sensor configured at a lighting node includes: Array photoelectric sensors configured at lighting nodes collect road illumination distribution parameters and simultaneously obtain power waveform characteristic parameters; Gridding the road illumination distribution parameters according to the geographical location information of the lighting nodes, and performing Fourier transform on the power waveform characteristic parameters to generate a correlation data set of a time-frequency characteristic matrix and a spatial illumination matrix; Cleaning abnormal data in the associated dataset using a dynamic threshold model of historical equipment operation data to generate a spatiotemporal fusion dataset; The spatiotemporal fusion dataset is input into a federated learning framework, and the missing data is interpolated and compensated through a parameter exchange protocol between edge computing nodes to generate a multi-source fusion dataset of all the lighting nodes.

3. The energy-saving optimization method for smart city lighting and energy management according to claim 1, characterized in that: The step of performing coupling analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data to determine the dynamic characteristic parameters of the lighting device includes: Performing multi-scale decomposition on the junction temperature fluctuation features in the multi-source fusion data to extract a junction temperature modal feature set; Perform energy distribution analysis on the harmonic spectrum of the power supply system to generate harmonic energy density spectrum characteristics; Generate a cross-domain coupling feature matrix by performing tensor space mapping on the time domain evolution law of the junction temperature modal feature set and the frequency domain distribution characteristics of the harmonic energy density spectrum feature; The transient response amplitude and trend change rate in the cross-domain coupling characteristic matrix are input into a nonlinear degradation mapping function to generate a dynamic characteristic parameter set.

4. The energy-saving optimization method for smart city lighting and energy management according to claim 3, characterized in that: The dynamic characteristic parameters include a light efficiency decay rate, a voltage sensitivity factor, and a thermal resistance dynamic parameter. The step of constructing a multi-objective optimization function based on the dynamic characteristic parameters includes: Constructing a mapping matrix of device performance degradation by performing interval clustering on the light efficiency decay rate, the voltage sensitivity factor, and the thermal resistance change curve; Performing state mapping on the joint state of the current device according to the change rate and stable boundary value of each sub-segment in the mapping matrix, and generating a parameter attribution vector that transforms the state label into the objective function variable; Generate a multi-objective coupling structure by performing time-series synchronization on the parameter attribution vector, the illumination demand matrix, the power load forecast data, and the historical abnormal response label; Performing weighted combination on each objective function sub-element in the multi-objective coupling structure to generate an objective priority assignment vector; According to the fluctuation sensitivity and frequency-related factors of each target dimension in the target priority assignment vector, the energy consumption term, response time term and illumination deviation term in the objective function are segmented and reconstructed to construct a multi-objective optimization function.

5. The energy-saving optimization method for smart city lighting and energy management according to claim 1, characterized in that: The step of processing the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy includes: Decomposing the multi-objective optimization function into a device-level dynamic optimization sub-function and a system-level steady-state optimization sub-function; Performing dynamic parameter space compression on the device-level dynamic optimization subfunction, iteratively shrinking the feasible domain of the device control parameters by dynamically adjusting the compression rate of the parameter search space, and generating a device-level dynamic optimization solution set; Performing distributed coupling constraint processing on the system-level steady-state optimization subfunction to generate a system-level steady-state optimization solution set; Multi-objective collaborative decision-making is performed on the device-level dynamic optimization solution set and the system-level steady-state optimization solution set, and the optimization conflicting objectives are prioritized based on the non-dominated solution screening mechanism to generate a multi-domain collaborative optimization control strategy including pulse width modulation instructions and power supply phase compensation parameters.

6. The energy-saving optimization method for smart city lighting and energy management according to claim 5, characterized in that: After the step of processing the multi-objective optimization function to generate a multi-domain collaborative optimization control strategy, the method further includes: Inputting the multi-domain collaborative optimization control strategy into the digital twin model; Performing spatiotemporal synchronization mapping on the pulse width modulation instruction and the power supply phase compensation parameter through a virtual execution interface to generate a virtual control instruction set; Dynamically modifying parameters of the virtual control instruction set based on multi-source fusion data collected in real time; The modified instruction set is subjected to co-simulation of illumination mutation and load fluctuation to generate a prediction response matrix. The power supply phase compensation parameters are iteratively weighted according to the predicted response matrix to generate a feedback correction control strategy.

7. The energy-saving optimization method for smart city lighting and energy management according to claim 5, characterized in that: The method further comprises: Constructing equipment health evaluation indicators based on the multi-domain collaborative optimization control strategy; Determining a maintenance priority sequence for each lighting node by performing spectral decomposition on the correlation between the equipment health evaluation index and the historical operating parameters; Performing distributed weighted mapping on the maintenance priority sequence to generate a predictive maintenance plan including equipment status warning indicators and maintenance time windows; The configuration parameters of the multimodal sensor are adaptively adjusted according to the predictive maintenance scheme to determine a dynamic allocation strategy for the sampling frequency and monitoring accuracy of the sensor.

8. An energy-saving optimization device for smart city lighting and energy management, characterized in that: The energy-saving optimization device for smart city lighting and energy management includes: An acquisition module is used to acquire multi-source fusion data collected in real time by multi-modal sensors configured at lighting nodes; an analysis module, configured to perform a coupling analysis on the junction temperature fluctuation characteristics of the lighting device and the harmonic spectrum of the power supply system based on the multi-source fusion data, and determine the dynamic characteristic parameters of the lighting device; A construction module, configured to construct a multi-objective optimization function according to the dynamic characteristic parameters; A generation module is used to process the multi-objective optimization function and generate a multi-domain collaborative optimization control strategy.

9. An electronic device, characterized in that: Comprising a memory and a processor, wherein: The processor is configured to execute a computer program stored in the memory; When the processor executes the computer program, it implements the steps in the energy-saving optimization method for smart city lighting and energy management as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the energy-saving optimization method for smart city lighting and energy management described in any one of claims 1 to 7 are implemented.

Images