A method for estimating the light flux decay state of a mixed-age luminaire network and for group control illumination equalization compensation

Abstract

The application discloses a kind of different age lamps network light flux recession state estimation and group control illumination equalization compensation method, belong to wisdom city road lighting control technical field.Aiming at the problem that multiple batches of LED street lamps formed by city road renovation in stages coexist in the same group control network and unified dimming leads to uneven road illumination, the application only uses the inherent electrical monitoring data of the driver, extracts the aging forward voltage drift by temperature-aging decoupling and maps it as the light flux maintenance rate estimate, solves the differentiated compensation dimming increment in combination with the road illumination contribution matrix, realizes sensorless self-calibration closed loop through adjacent pole power consumption statistics mutual school, supports the rolling expansion of staged renovation scene, and can make the road illumination uniformity continuously meet the design requirements.

Description

Technical Field

[0001] This invention belongs to the field of smart city road lighting control technology, specifically relating to a method for estimating the luminous flux decay state of luminaire networks of different ages and compensating for the illuminance balance of group control. Background Technology

[0002] With the continuous advancement of urban road lighting energy-saving renovation projects, LED streetlights have become the mainstream type of urban road lighting. However, due to constraints in construction funding and construction conditions, road lighting renovations are often implemented in phases. Different sections of the same road are replaced with lights in batches in different years, resulting in a typical engineering scenario where multiple batches of LED lights with different service lives coexist on the same group control network for a long time.

[0003] During long-term operation, LED luminaires experience a continuous decline in luminous flux due to aging mechanisms such as the accumulation of junction defects and an increase in the proportion of non-radiative recombination. When multiple batches of luminaires with installation ages differing by several years coexist in the same group control circuit, the group control system's issuance of uniform dimming commands to all luminaires will result in significant unevenness in road illuminance. Luminaires with longer service lives, due to severe luminous flux degradation, will produce significantly lower actual road illuminance under the same drive level compared to newly installed luminaires, creating visible illuminance bands along the road's longitudinal direction. This violates the current "Urban Road Lighting Design Standard" (CJJ45) regarding road illuminance uniformity. The mandatory requirements have an adverse impact on road driving safety and pedestrian visual environment.

[0004] To address the aforementioned issues, the existing technology has the following shortcomings.

[0005] First, consider installing road surface photometric sensors. This involves installing illuminance sensors on light poles or along the road surface to directly measure actual light output. However, this approach suffers from high costs associated with purchasing and installing individual lamp sensors, difficulty in guaranteeing long-term reliability during outdoor operation, and limited feasibility for large-scale, dense deployment projects, thus restricting its widespread application in major road lighting renovation projects.

[0006] Second, a complete replacement plan. Replacing all streetlights with new ones from the same batch would eliminate batch differences, but mandatory scrapping would still be a factor. In-service equipment during its lifespan results in a waste of equipment resources and requires a large one-time capital investment, making it difficult to implement when the city's lighting authorities have limited funds.

[0007] Third, a luminous flux decay prediction scheme based on the IESLM-80 / TM-21 standard. This scheme utilizes long-term luminous flux decay test data before the luminaire leaves the factory, and predicts the luminous flux maintenance rate of in-service luminaires based on the TM-21 extrapolation model, thereby performing differentiated dimming compensation. However, the TM-21 method is designed for new product factory testing, and its prediction accuracy depends on a complete understanding of the luminaire's entire thermal history. For in-service luminaires that have been operating outdoors for many years, with unknown actual thermal histories and batch-to-batch individual differences, the prediction error introduced by TM-21 extrapolation usually cannot meet the estimation accuracy requirements for illuminance uniformity compensation.

[0008] Furthermore, although existing patent literature has proposed the idea of ​​estimating the luminous flux of a single lamp using the forward voltage drift of an LED, this solution is only applicable to single-lamp closed-loop control in a laboratory environment. It does not involve multi-lamp network-level collaborative compensation, nor does it involve outdoor in-service scenarios with unknown thermal history. It also does not consider the problem of group control and scheduling of lamps of different ages unique to the phased renovation of urban roads. Therefore, it is fundamentally different from the technical problem addressed by this invention in terms of scenario dimension.

[0009] In summary, the existing technology lacks a complete technical solution that can non-intrusively estimate the luminous flux decay status of outdoor LED street light networks of different ages using only the inherent electrical monitoring data of the lamp driver, without adding new photometric sensors or forcibly replacing equipment, and thereby achieve balanced compensation for road surface illuminance uniformity within the group control framework. Summary of the Invention

[0010] To address the aforementioned problems, this invention proposes a method for estimating the luminous flux decay state of luminaire networks of different ages and for compensating for the illuminance balance of group control, comprising the following steps:

[0011] The method, applicable to group-controlled lighting networks formed by phased urban road reconstruction and comprising several LED streetlights installed in different batches, includes the following steps:

[0012] S1, Electrical reference fingerprinting: When each luminaire is connected to the group control network, the data of that luminaire under several preset driving conditions is collected. The initial positive voltage response under the condition of driving current. With the temperature of the driver housing Constructing an initial positive voltage reference fingerprint for a two-dimensional index and install timestamp Batch identification and rated initial luminous flux Store them together in the lighting fixture archive;

[0013] S2, In-service electrical data periodic acquisition: Read the current forward voltage from each lamp driver according to the set acquisition period. Drive current and driver housing temperature ;

[0014] S3, Decoupling junction temperature estimation from temperature-aging component: with Estimate the current junction temperature of the luminaire and extract the forward voltage drift caused solely by aging. :

[0015] ;

[0016] in The forward voltage drop temperature coefficient of the LED junction region. For LED junction-to-case thermal resistance, The reference junction temperature used when establishing the baseline data;

[0017] S4, Non-invasive estimation of luminous flux maintenance rate: The luminous flux maintenance rate of each luminaire is estimated by converting the values ​​through a calibration mapping function. :

[0018] ;

[0019] Where the mapping parameters , Obtained by the self-calibration procedure in step S7 or by regression calibration of a subset of lamps with known ages;

[0020] S5, Construction of the Road Illumination Contribution Matrix: Based on the installation coordinates, pole height, and light distribution curve of each luminaire, a luminaire-road illuminance contribution coefficient matrix is ​​constructed. ,element Indicates the first The light pole is the first The theoretical illuminance contribution value of each road surface evaluation point under unit luminous flux conditions. The total number of road surface evaluation points. This represents the total number of light poles.

[0021] S6, Differentiated compensation dimming increment solution and distribution: constrained by road surface illuminance uniformity Given the constraints of the luminaire dimming range, and with the optimization objective of minimizing total power consumption, we solve for the individualized dimming compensation increment vector for each luminaire. :

[0022] ;

[0023] in The unified baseline dimming command issued by the group control system will be solved to generate individualized dimming commands. The signal is sent to the corresponding lighting fixture driver via unicast.

[0024] S7, Mutual calibration and self-verification of power consumption statistics between adjacent poles: During the effective nighttime period when weather conditions are stable, the power consumption statistics of each lamp in the same group control section are verified under group control commands. Actual power consumption Perform statistical distribution analysis, using the median power consumption of the segment. As a reference benchmark, the power consumption deviation is calculated to exceed the threshold. luminous flux maintenance correction of luminaires The feedback is then used to update the estimated value in step S4, achieving a system-level self-calibration closed loop without an external photometric reference sensor. The corrected value is then updated. Step S6 is calculated for the next compensation cycle.

[0025] In step S3, when the luminaire's factory data sheet does not provide thermal resistance... At that time, a multi-condition thermal balance identification procedure was used to determine the initial installation stage. The luminaire was operated to thermal steady state under at least three different power conditions, and the corresponding data were collected. Solve the following overdetermined linear equation system simultaneously using the least squares method with a set of three variables, and simultaneously identify... and :

[0026] ;

[0027] Identification results and Stored in the lighting equipment archive for direct retrieval during subsequent service.

[0028] In step S4, the mapping parameters and The calibration adopts a hierarchical regression strategy: for a subset of lights with clearly defined installation timestamps within the group control network, its... The measured sequence is used as the input feature, and the theoretical luminous flux maintenance rate is calculated based on the IESTM-21 extrapolation model at the corresponding cumulative lighting hours. To supervise the annotation process, power-law model parameters were fitted using the weighted least squares method, and parameter consistency constraints were applied to different batches of the same lamp model; for each lamp... Output synchronously calculate confidence scores The confidence score is determined by The monotonicity of the time series, the noise level, and its deviation from the mean of the same batch are jointly determined when... When the value falls below the preset confidence threshold, a manual review flag is sent to the operations and maintenance platform.

[0029] In step S5, the elements of the illuminance contribution coefficient matrix Calculated using the following optical geometric model:

[0030] ;

[0031] in For the first The light pole is pointing towards the road surface evaluation point. Luminous intensity in the direction (obtained by interpolation from the light distribution data). For the first The installation height of the light pole, This is the angle of incidence of light relative to the road surface normal. When measured light distribution data for the luminaire is unavailable, a standard light distribution curve template matching the luminaire's model classification code is used for calculation, and this column is marked as the template estimate in the file. This is automatically triggered when measured light distribution data is added. Local refinement updates.

[0032] In step S6, the actual illuminance at each evaluation point on the road surface is calculated using the following dimming-luminous flux coupling model:

[0033] ;

[0034] in This is a mapping function from dimming commands to output luminous flux, and it determines the PWM dimming mode. For constant current analog dimming method ,index Determined by the type of drive; pavement illuminance uniformity is constrained by a minimum-to-mean ratio:

[0035] ;

[0036] Number of lampposts When the value is small, the optimization problem is solved by relaxing linear programming. When it is larger, use The grouped iterative greedy algorithm obtains the suboptimal solution, and the optimization result is cached in the controller's local storage, with its validity period extending to the next calibration cycle.

[0037] When the number of light poles in the group control section When the set threshold is exceeded or the controller's computing resources are limited, the solution in step S6 replaces the global optimization with the following hierarchical approximation strategy: the luminaires are approximated based on the estimated luminous flux maintenance rate. Divided into recession level Lighting fixtures of the same grade share the same compensation increment. The compensation increment for each level is calculated using the following formula:

[0038] ;

[0039] in This is the weighted average of the luminous flux maintenance rate of all luminaires in the network. For the first Lighting fixtures within the specified grade The average grade This is an asymptotic compensation coefficient used to suppress the illuminance inversion phenomenon caused by discretization at the boundary of adjacent levels.

[0040] In step S7, the adjacent pole power consumption statistics mutual calibration procedure is specifically executed as follows: based on the change in ambient temperature within a time period... Furthermore, the absence of precipitation events serves as a valid calibration window criterion; within the valid window, the power consumption of each lamp in the same group control section is evaluated. Fitting a normal statistical distribution ; for satisfying For abnormal luminaires, the luminous flux maintenance correction is calculated using the following formula:

[0041] ;

[0042] in The power consumption-luminous efficacy correlation coefficient was determined statistically from measured data of the same model of lamps; the corrected value is... Through the forgetting factor An exponentially weighted moving average filter is incorporated into the historical estimation sequence to suppress estimation abrupt changes caused by occasional interference.

[0043] For historical in-service luminaires without initial baseline fingerprint records, the method further includes a cold start calibration procedure: during the first maintenance power outage restart phase after the initial system deployment, when the luminaire junction temperature has returned to ambient temperature, the cold-state forward voltage is collected at this time. As a pseudo-benchmark; combined with the statistical prior distribution of the factory luminous flux maintenance rate corresponding to the lamp model, and the known age of lamps in the same area. Statistical characteristics, inferred from Bayesian methods, indicate historically in-service lighting fixtures. An initial posterior estimate is assigned; as subsequent self-calibrated data accumulates, the posterior estimate gradually converges to a value based on... Direct estimates of physical mappings.

[0044] When a new batch of lighting fixtures is connected to the group control network, the method also automatically triggers the phased upgrade batch access process: collecting and storing the initial baseline fingerprint of the new batch of lighting fixtures. The matrix corresponding to the road surface evaluation points within the influence range of the new lighting fixtures and adjacent lighting fixtures. Incremental reconstruction of rows and columns is performed to avoid full matrix recalculation; the compensation incremental optimization of step S6 is re-executed and the issued instructions are updated for the road sections affected by the new batch; the new batch of lamps are included in the self-calibration participation set of step S7, and participate in the whole network mutual calibration after at least one complete calibration window, and the normal dimming operation of existing lamps is not interrupted during the entire access process.

[0045] This method supports switching between two compensation scheduling modes: in periodic static compensation mode, batch updates are performed on a daily basis. With compensation increment vector Within the same day, a fixed compensation increment is used; in the event-triggered dynamic compensation mode, when the self-calibration module in step S7 detects that the power consumption deviation of a certain lamp exceeds the dynamic threshold, or when the new batch access event described in step S9 occurs, the local segment is immediately triggered. Revaluation and The optimization process ensures that the new compensation command is issued within the next dimming control cycle after the event occurs.

[0046] Compared with the prior art, the present invention has at least the following beneficial effects.

[0047] 1. No additional photometric sensors are required. This invention utilizes only the inherent forward voltage, drive current, and case temperature measurement functions of the LED street light driver as the sole source of information. It achieves non-intrusive estimation of luminous flux decay without adding any additional hardware sensors, significantly reducing system modification costs and maintenance complexity, and demonstrating strong engineering economics.

[0048] II. Effectively addressing the problem of unknown thermal history. This invention uses a benchmark fingerprinting and temperature-aging decoupling mechanism to separate and extract the systematic drift of positive voltage caused by aging from the temperature-related components. This overcomes the dependence of existing prediction methods such as TM-21 on complete thermal history records and can be directly applied to in-service lighting fixtures that have been operating outdoors for many years and lack thermal history information.

[0049] Third, achieving a complete closed loop from single-lamp estimation to network-level illuminance equalization. This invention combines single-lamp lamp attenuation state estimation with the road illuminance contribution matrix, using road illuminance uniformity as a constraint and minimizing total power consumption as the objective. By solving the differential compensation dimming increment through constrained optimization, a complete technical closed loop from electrical signal sensing to road illuminance equalization control is formed, overcoming the limitation of existing single-lamp estimation schemes that do not involve network-level coordination.

[0050] IV. Naturally supports dynamic expansion for phased renovation scenarios. This invention designs an expansion access mechanism based on batch filing and incremental matrix updates. When a new batch of lighting fixtures is connected, only the calculation parameters of the affected sections need to be updated locally, without the need to rebuild the system state globally, thus achieving seamless adaptation and smooth expansion for phased renovation scenarios.

[0051] Fifth, it possesses the system self-calibration capability without reference sensors. The adjacent pole power consumption statistical mutual calibration mechanism of this invention uses the consistency of power consumption statistics among lamps in the same section under the same dimming conditions as an internal reference benchmark. It can periodically correct the light attenuation estimation results without any external reference photometric sensor, thereby improving the long-term stability of the estimation accuracy.

[0052] VI. Compensation Optimization Balances Illumination Uniformity and Energy Saving Targets. This invention takes minimizing total power consumption as the optimization target and achieving illuminance uniformity as the constraint. This allows differentiated compensation to meet the CJJ45 illuminance uniformity requirements while achieving overall energy saving by appropriately reducing the drive level of lamps with less degradation, thus avoiding the energy waste caused by raising all lamps to the highest compensation level. Attached Figure Description

[0053] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative effort. In the drawings:

[0054] Figure 1 This is a complete flowchart of the method described in this invention;

[0055] Figure 2 This is a schematic diagram illustrating the calibration mapping relationship between the positive voltage drift during temperature-decoupled aging and the luminous flux maintenance rate. Detailed Implementation

[0056] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0057] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, 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 application pertains.

[0058] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0059] Example 1

[0060] This embodiment aims to demonstrate the establishment of the system's physical foundation layer—without adding any new road surface photometric sensors, initial electrical baseline fingerprints were collected and stored for three batches of luminaires of different ages using a phased filing method, and a reliable in-service electrical data periodic acquisition mechanism was configured. The baseline fingerprints and in-service real-time data together constitute the subsequent light decay estimation.

[0061] Taking a phased lighting renovation project on a main urban road (two-way four lanes, approximately 22 meters wide) as an example, the specific implementation of this invention is illustrated. This road section is approximately 800 meters long. LED streetlights are arranged crisscrossingly along both sides of the road according to CJJ45 regulations, with a pole spacing of approximately 17 meters, totaling 48 poles, numbered sequentially as follows: to .

[0062] This section of road has undergone three phases of renovation. The installation times and coverage areas for each phase are as follows: Phase 1 (times) )Finish to The second phase (time) involves replacing the lights on 24 lampposts in the northern section. ,distance Completed in approximately 36 months to A total of 20 light poles in the southern section; Phase III (Time) ,distance Completed in approximately 60 months to Four light poles were installed at the intersections. The three batches of light fixtures have the same nominal model and rated initial luminous flux. All are nominally rated at 12000lm (the following parameters are all example values, the same below), with a rated drive current of 12000lm. All three batches of products are 700mA, but they come from different production cycles, and there are batch-to-batch differences in chip batches, phosphor formulations, and packaging processes.

[0063] exist After the entire system is fully deployed, the cumulative lighting time of the three batches of lights (estimated based on 12 hours of operation per day) is approximately as follows: First batch Hours, second batch Hours, third batch (New Installation). If the group control system uniformly issues a 70% dimming command to all 48 light poles along the entire road section, the actual road surface illuminance output of the three batches of lights will significantly differ due to variations in luminous flux decay. Based on theoretical estimations using the IESTM-21 extrapolation model, under the aforementioned cumulative lighting duration, the luminous flux maintenance rate of the first batch of lights is approximately... The range (specific values ​​depend on the chip model and driving thermal environment, the same below), the second batch is approximately in The third batch has an illuminance of 1.00. The difference in illuminance between the three batches is sufficient to create visible light and dark bands on the road surface, violating CJJ45's requirement for uniform road surface illuminance. Requirements.

[0064] This invention addresses the aforementioned scenario by utilizing the existing electrical measurement capabilities of each luminaire driver to achieve non-invasive estimation and differentiated dimming equalization compensation of the luminous flux decay state of three batches of luminaires without adding any new road surface photometric sensors.

[0065] The system involved in this embodiment consists of the following hardware layers.

[0066] Each light pole has a built-in intelligent driver, belonging to the lighting fixture layer, and is an intelligent LED driver. It supports DALI protocol communication and has the following electrical measurement capabilities: forward voltage. The measurement resolution is no less than 1mV and the accuracy is better than mV; Drive current The measurement accuracy is better than that of full scale. Driver PCB board temperature The measurement accuracy is better than 1°C. The above accuracy requirements are exemplary technical specifications and can be adjusted according to the specific device selection during actual implementation. The driver supports receiving unicast DALI individualized dimming commands with a dimming resolution of not less than 1%, and can operate concurrently with broadcast group control command channels.

[0067] Each centralized controller manages several light poles via the DALI bus (each controller manages 8 poles in this embodiment, for a total of 6 poles across the entire road section). It belongs to the aggregation layer and is the centralized controller. It is responsible for aggregating the lights under its jurisdiction. Telemetry data is transmitted to the edge controller via the upper-layer wireless link; at the same time, compensation instruction packets issued by the edge controller are received, unpacked according to the light pole address, and distributed to the corresponding driver via DALI unicast frames.

[0068] The group-controlled illuminance equalization compensation controller (edge ​​controller) belongs to the edge layer. One unit is deployed across the entire road segment, running all the algorithm modules described in this invention: temperature-aging decoupling and light decay estimation module, illuminance equalization optimization module, individualized command issuance module, and self-calibration and file management module. It is equipped with local non-volatile storage (capacity requirement approximately equal to the lamp archive plus historical time-series data, eMMC or SSD media optional), and communicates bidirectionally with six centralized controllers via an NB-IoT wide area network. Optimization calculations are performed locally, without relying on the real-time response of the upper-level urban lighting management platform.

[0069] The communication topology is explained. The underlying DALI bus carries the uplink of electrical telemetry data for the lighting fixtures and the downlink of dimming commands; the upper-layer NB-IoT link carries the data exchange between the central controller and the edge controller. Group control broadcasts dimming commands. With individualized compensation unicast instructions At the underlying level, it is sent in DALI broadcast frames and unicast frames respectively. The unicast frame header contains the target light pole address and the timeliness verification field to prevent the expiration compensation instruction from being executed incorrectly.

[0070] Step S1 is executed immediately after each batch of lamps has been installed, debugged, and stabilized upon initial power-on. It forms the data basis for all subsequent estimations and compensation calculations.

[0071] Lighting fixtures with rated current After power-on, the temperature of the driver PCB board... The ambient temperature gradually rises and then stabilizes. The edge controller reads data at 1-minute sampling intervals. When the maximum difference of 5 consecutive readings does not exceed When the luminaire has reached thermal stability, the fingerprint acquisition process is triggered. In one embodiment, the thermal stability waiting time is typically 20 to 40 minutes, depending on the luminaire's power rating and the heat dissipation conditions of the installation environment. The above-mentioned determination threshold and example values ​​of waiting time do not constitute a limitation on the scope of protection of this invention, and can be adjusted according to the actual characteristics of the luminaire during implementation.

[0072] After confirming thermal steady state, apply heat to the target luminaires sequentially. The system presets driving conditions and reads steady-state electrical data triplets under each condition. , .

[0073] In a preferred embodiment, the preset drive conditions cover three current levels: ;

[0074] The case temperature at each current level corresponds to 50%, 75%, and 100% of the rated current. The current ambient temperature and steady-state self-heating under this operating condition are jointly determined. The real-time value of the temperature sensor built into the driver is directly read, without the need for an externally controlled temperature environment. This feature allows the data creation program to be executed in situ in a normal outdoor environment after the luminaire is installed. After each operating condition switch, a wait of no less than 120 seconds is required to ensure thermal stability under the new operating condition. The edge controller reads the steady-state electrical data after the wait is over.

[0075] It should be noted that the selection of the three current levels mentioned above is only an example, and the number of operating conditions is not specified. The selection of [a specific parameter] should meet the requirements of subsequent thermal resistance identification and fingerprint interpolation accuracy. Typically... This should meet the requirements. For luminaires with a wide dimming range driven by the driver, the current setting can be appropriately increased to improve the fingerprint interpolation accuracy.

[0076] Finish After the group's operating conditions are collected, the edge controller uses... and Using two-dimensional coordinates, a two-dimensional lookup table covering the normal operating range of the luminaire is constructed through bilinear interpolation, forming an initial positive voltage reference fingerprint. .

[0077] In one embodiment, the resolution of the lookup table can be set to: 25mA increments for the current dimension (covering...). (Range), temperature range in 5°C increments (coverage) Scope, of which , The historical extreme temperatures at the installation location, (For full load self-heating), the table occupies approximately several KB of storage space, which does not put significant pressure on the edge controller's storage.

[0078] For each light fixture The edge controller writes the following fields into the lamp archive to form the initial file record for that lamp: Pole_ID: Pole number ( to );

[0079] Batch_ID: Batch identifier (in this embodiment, it is Batch-A, Batch-B, and Batch-C, which correspond to the three phases of renovation respectively).

[0080] Installation timestamp; Rated initial luminous flux, taken from the luminaire's factory data sheet; Rated drive current; Initial forward voltage reference fingerprint lookup table; Reference shell temperature at the time of filing (sequence of measured values ​​under various operating conditions); Reference junction temperature during baseline data establishment; Junction-shell thermal resistance identification value;

[0081] The aforementioned archive data is persistently stored on the edge controller's local non-volatile memory and is periodically backed up to the upper-level city lighting management platform, supporting full lifecycle traceability.

[0082] Because the three batches of lamps were respectively located at , , The filing was completed at three different times, and the baseline fingerprints for each batch were... and Keys are stored independently and do not interfere with each other. In the third batch of lighting fixtures ( to )At Once the archiving is completed, the lamp archive contains a complete record of the baseline fingerprints for all 48 lamp poles along the entire road. Subsequent steps S3 to S7 can perform light attenuation estimation and compensation optimization on the three batches of lamps in a unified and parallel manner, regardless of the order in which the batches were archived. This design gives the system inherent openness, allowing it to expand rollingly as the phased renovation progresses, without requiring all batches of lamps to be connected simultaneously.

[0083] After the baseline data is established, the edge controller initiates a periodic electrical data acquisition task for each lamp to continuously provide raw input for the aging status estimation in step S3.

[0084] In one embodiment, the data collection cycle can be set to once per calendar day, with the collection time window selected between 1:00 AM and 3:00 AM, during the period when the lighting fixtures are operating in a normal steady state. In another embodiment, for newly installed lighting fixtures with slow light decay, the data collection cycle can be appropriately extended to once every 5 to 7 days to reduce communication resource consumption; for lighting fixtures with a cumulative operating time exceeding [a certain duration], [the cycle can be extended to once every 5 to 7 days]. For aging luminaires with 70% of their lifespan remaining, the data acquisition cycle can be shortened to once every 12 hours to improve status tracking accuracy. Optionally, the data acquisition task can be decoupled from the luminaire's daily dimming operation cycle, triggering only after the luminaire has been in the set dimming level for more than 30 minutes, to ensure that the luminaire is in a thermally stable state during data acquisition.

[0085] The prerequisites for triggering a valid data acquisition are: the lights have been running continuously for at least 30 minutes; and the current group control dimming command is active. The data acquisition window remains stable; the driver has no fault alarm flags. Once these conditions are met, the central controller reads the telemetry registers of each lamp driver under its jurisdiction and sends the data packet... The data is reported to the edge controller via NB-IoT.

[0086] The edge controller performs the following compliance checks on each reported data entry: It must be within the normal operating voltage range of the luminaire, which can be set according to the luminaire specification sheet. For example, for a single-chip module with a rated voltage of 700mA, the compliance range can be set to... V, this value is for example only; Must be consistent with the current dimming command The corresponding expected current matches the set tolerance; Data points must be within a reasonable physical range. Data points that do not meet any of the above conditions are marked as invalid and skipped, and will not participate in subsequent aging estimation calculations; if multiple invalid data points occur consecutively, a driver abnormality alarm will be triggered, notifying the operations and maintenance platform for manual inspection.

[0087] Valid data is written to the edge controller's local time-series database in separate tables according to the light pole number, and each record contains a timestamp. The dimming instructions at that time and triplet .

[0088] Step S2: Acquisition and storage of in-service electrical data triplets It is the direct input for the temperature-aging decoupling calculation in step S3. Its core logic lies in: for the same light fixture, the baseline fingerprint established in step S1... The voltage response baseline of the luminaire under various operating conditions was recorded at its initial luminous efficacy state at the factory; step S2 then continuously collects the real-time voltage response under the same operating conditions during its service life. The difference between the two, after thermal drift caused by changes in the stripping junction temperature, constitutes a systematic positive voltage shift purely caused by aging. It is this difference—not... The absolute value itself—carrying physical information about the LED luminous flux decay state—constitutes the core observable of the non-invasive light decay estimation mechanism of this invention. The above data acquisition and storage process is automatically scheduled by the edge controller, requiring no manual intervention.

[0089] Example 2

[0090] This embodiment, based on the electrical reference fingerprint established in Embodiment 1 and the in-service data acquisition, details the specific implementation of steps S3 and S4, namely, how to separate the positive voltage drift caused by pure aging from the driver electrical monitoring data. And map it to an estimate of the luminous flux maintenance rate for each luminaire. The parameter identification and calibration procedures involved provide specific support.

[0091] Before formally introducing the calculation process, it is necessary to explain This serves as the physical basis for observable luminous flux decay.

[0092] LED forward voltage Under constant current driving conditions, the following two types of effects prevail: one is the thermal effect, where the forward voltage drop of the LED junction increases with the junction temperature. The temperature coefficient increases and decreases linearly. For InGaN-based white LEDs, they are typically in the... to interval (with) m (These are typical example values; the same applies below. Actual values ​​may vary depending on the chip batch.) Secondly, there is the aging effect. As the cumulative operating time of the LED increases, defects within the junction region (dislocation multiplication, non-radiative recombination centers formed by metal diffusion) gradually accumulate, increasing the proportion of non-radiative recombination of charge carriers. This process, while altering quantum efficiency (leading to luminous flux attenuation), also causes systematic changes in the junction's equivalent series resistance and ideality factor, thus affecting the performance of the same LED. Under certain conditions, a measurable systematic drift in the positive voltage is generated. .

[0093] The key difference lies in the fundamental difference in time scale between the two types of effects: thermal effects exhibit transient responses to changes in operating conditions within seconds; while aging effects accumulate slowly over hours to months, and can be considered quasi-static quantities within a single acquisition window. This separation of time scales provides the physical premise for decoupling the two types of effects in electrical data. This invention utilizes the benchmark fingerprint established in step S1 as an "aging-free reference baseline," and through controlled comparison with in-service measured values, separates and extracts the slowly changing aging drift signal from the rapidly changing thermal response signal.

[0094] It should be noted that for LED lamps of different models and with different packaging structures, The correlation strength between the luminous flux maintenance rate and the luminous flux maintenance rate may vary. This invention does not require that the correlation be consistent across all luminaires, but rather establishes a mapping relationship independently for each luminaire model through the statistical calibration procedure in step S4, in order to accommodate the differentiated characteristics of different products.

[0095] Junction-shell thermal resistance It is a key parameter for estimating junction temperature. In this invention, there are two ways to obtain it, which can be flexibly selected according to the availability of lamp data.

[0096] For example, data can be directly retrieved from the lighting fixture's manufacturer's data sheet, or multi-condition thermal balance can be identified in situ when the lighting fixture's data sheet does not provide this information. If it is necessary to verify the thermal resistance in situ after installation to improve the estimation accuracy, the following identification procedure shall be executed simultaneously during the baseline documentation stage in step S1.

[0097] Make the lighting fixtures ( The data were collected from the steady-state electrical ternary arrays under different power conditions until thermal steady-state operation. Under steady-state thermal conditions, all the power consumption of the LED chip is converted into heat, and the junction-case temperature difference satisfies the steady-state thermal circuit equation:

[0098] ;

[0099] Meanwhile, the forward voltage of the same lamp at different junction temperatures follows a linear temperature dependence:

[0100] ;

[0101] Expanding the above equation and substituting it, we get: and Equations with unknowns:

[0102] ;

[0103] right The collected data were used to construct a system of overdetermined linear equations, which were then solved simultaneously using the weighted least squares method. and In one embodiment, weights can be assigned based on the signal-to-noise ratio (SNR) of the data collected under each operating condition, with lower SNR conditions (such as low current and low temperature rise conditions) assigned smaller weights. (The identified data is then processed.) and The information is written into the lighting fixture database and can be directly retrieved during subsequent service without repeated identification, unless the lighting fixture is replaced or the heat dissipation structure is changed.

[0104] Taking the specific scenario of this embodiment as an example, in the system at When initially deployed, for to All 48 light poles will undergo the above identification procedure. The identification process can be combined with the baseline fingerprint collection in step S1 and completed in the same debugging process without adding extra downtime.

[0105] The edge controller collects valid data each time. After obtaining the data triples, immediately perform the following junction temperature estimation calculation.

[0106] Current junction temperature estimate: ;

[0107] in This is the real-time power consumption of the LED at the current sampling moment. , The identification or lookup value of the lamp is taken from the lamp archive. It is calculated directly from the measured values ​​reported by the driver and the parameters in the database, without the need for any additional sensors.

[0108] Reference junction temperature calculation: Similarly, for each reference operating condition of this luminaire during step S1 documentation, the reference junction temperature is calculated using the same thermal resistance model:

[0109] ;

[0110] Take the calculated value under the reference working condition as Stored in the lighting fixture archive.

[0111] The decoupling computation is performed in two steps. Step 1: Baseline fingerprint query. Based on the fingerprint collected at the time... Using an index, the reference forward voltage is obtained by interpolation in the two-dimensional lookup table established in step S1. This value represents the expected forward voltage of the luminaire under initial luminous efficacy and current identical external conditions. Step 2: Aging drift extraction. Calculate the forward voltage drift and correct for the second-order junction temperature deviation caused by changes in self-heating power dissipation.

[0112] ;

[0113] The corrections need to be explained here. The physical meaning: When a lamp causes... When changes occur, at the same shell temperature LED real-time power consumption This also causes slight changes, resulting in a decrease in the junction temperature. With reference junction temperature A second-order bias is generated between them. Without correction, this bias will be transmitted through thermal effects. The introduction of spurious contributions interferes with the extraction of aging drift; the introduction of the correction term eliminates this coupling, making... This becomes a more objective measure of the pure aging state. In one embodiment, for LED luminaires of typical street lighting power levels, this correction term is on the order of approximately to mV, relative to aging-induced (Typically in the range of several millivolts to tens of millivolts) It is a second-order small quantity, but it cannot be ignored in high-precision estimation scenarios.

[0114] Each valid data acquisition is calculated independently and appended to the historical time-series database of the luminaire, forming a monotonic time-series that increases with the cumulative operating time (for luminaires exhibiting...). The model of the lamp increases with age. It should be non-negative and increase slowly over time; for If the aging decreases, it is non-positive; the specific sign is determined by the aging characteristics of the lamp model, and the calibration procedure for the mapping parameters applies to both cases).

[0115] get Then, the edge controller calculates the luminous flux maintenance estimate for each luminaire using the following calibration mapping function:

[0116] ;

[0117] exist In a typical case, the above equation simplifies to the standard form in step S4. Mapping parameters , Physical constraints must be satisfied: for any expected... All possible values ​​are available. ,and hour (New lighting status).

[0118] The power-law model was chosen based on the following considerations: First, it has few parameters (only two), making it easy to fit robustly to a limited set of calibration samples; second, the exponential... The adjustable concavity and convexity of the mapping curve can adapt to different decay mechanisms. - The nonlinear form of the relationship; thirdly, the power-law model is consistent with the power-law lifetime prediction framework commonly used in LED aging kinetics, facilitating verification with the extrapolation results of TM-21. The selection of the above-mentioned functional form does not constitute a limitation on the scope of protection of this invention.

[0119] Mapping parameters and The following hierarchical regression calibration procedure is used to obtain the calibration, which is executed locally on the edge controller without requiring a network connection to an external computing platform.

[0120] Among the 48 light poles along the entire road section, the batch identifier (Batch_ID) and installation timestamp are used. The specific lighting fixtures and their cumulative hours of operation. (in The average daily lighting duration (which can be accurately integrated from historical dimming records) is a known quantity and can be used as a calibration sample set. In the scenario of this embodiment, The cumulative durations of the three batches of lamps are approximately 43,800 hours, 14,400 hours, and approximately 0 hours, respectively. The age range covered by the three batches is sufficient to support regression calibration.

[0121] For each luminaire of known age, the actual cumulative lighting hours are calculated using the IESTM-21-11 extrapolation model. Calculation of theoretical luminous flux maintenance rate The initial decay curve parameters required for TM-21 extrapolation are read from the LM-80 test report of the same type of luminaire; when the LM-80 report is unavailable, the industry statistical prior value of the same type of luminaire can be used instead, and marked in the confidence score. It should be noted that this invention does not require the theoretical prediction of TM-21 to have absolute accuracy, but rather uses it as a relative ranking benchmark in the calibration regression. The final estimation accuracy of the system will be further corrected and calibrated by the adjacent bar mutual calibration self-calibration in step S7.

[0122] Based on each calibrated sample lamp The measured mean is the feature input, with To label the target, the following objective function is minimized using weighted least squares:

[0123] ;

[0124] in To calibrate the sample set, weights Proportional to each sample For signal-to-noise ratio and TM-21 prediction confidence levels, samples with larger cumulative hours (more clearly defined aging conditions) are given higher weights. Optionally, soft constraints on parameter consistency are applied to different batches of the same lamp model.

[0125] ;

[0126] Constraint strength parameters , With tolerance It can be set based on prior knowledge of batch differences; when the source of chips of the same model from different batches is clearly consistent, stronger constraints can be applied; when the batch differences are large, the constraints can be appropriately relaxed.

[0127] In one embodiment, The reasonable range of values ​​is approximately (by (when the unit is mV) The reasonable range of values ​​is approximately The above range is for reference only. The actual calibration values ​​for different lamp models may exceed the above range. The actual calibration results should be used in practice. The above values ​​do not constitute a limitation on the scope of protection of this invention.

[0128] After calibration, and Perform physical plausibility verification: based on observations within the current network. Substitute the maximum value into the mapping function and verify. (This range corresponds to the decay range typically covered by LM-80; the lower limit can be adjusted according to the lamp model.) If the verification fails, the parameter abnormality flag is triggered and the calibration value of the previous cycle is retained, while an alarm is sent to the operation and maintenance platform.

[0129] parameter and It is not a fixed constant, but is updated periodically as the system accumulates more in-service data. In one embodiment, the calibration parameters can be updated every 90 days; in the early stages of system deployment when the calibration sample size is small, the update cycle can be appropriately shortened; when the age range of each batch of lamps is small, the update cycle can be extended to reduce unnecessary refitting.

[0130] For each light fixture each time Output: The edge controller synchronously calculates the confidence score. , The larger the value, the more reliable the estimation result. The scores are weighted and combined from the following three sub-scores:

[0131] Among them, the monotonicity component Examining history Does the time series exhibit a monotonic trend consistent with physical expectations? Specifically, the monotonicity of the time series is quantified using the Spearman rank correlation coefficient, and for correlation coefficients close to... The lighting fixtures give high scores Light fixtures with significantly non-monotonic or oscillating timing sequences are assigned low scores to identify fixtures that may have driver measurement errors or poor contact. Noise level component : From history The inverse of the coefficient of variation (the ratio of standard deviation to mean) of a time series signal-to-noise ratio is a measure of its strength; the lower the coefficient of variation, the lower the signal-to-noise ratio. Higher. Batch deviation items Calculate the light fixture's The standardized deviation of the luminaires compared to the mean of the same batch (same Batch_ID and similar cumulative duration) exceeds the standard deviation of the same batch. When doubled, Linear decrease. In one embodiment, The value can be 2.0 to 3.0, and the above values ​​do not constitute a limitation on the present invention.

[0132] Overall score Weight Optional As the default configuration. When When the level falls below a preset confidence threshold, a manual review flag is sent to the operations and maintenance platform; this lighting fixture's... The estimated value is still included in the compensation calculation in step S6, but the range of the compensation increment is constrained to a more conservative range to prevent overcompensation caused by low-confidence estimates.

[0133] by Taking the status of the entire road section system after several data collection cycles as an example, this illustrates the differentiated performance of the above steps on three batches of lamps of different ages.

[0134] The first batch of lighting fixtures ( to (Accumulated over approximately 43,800 hours), due to long-term aging, its positive voltage reference offset... It should be significantly greater than zero, corresponding to The estimated value will be lower than that of the second and third batches of luminaires. In one exemplary description, if the batch of luminaires... If it is in the range of several millivolts, then the mapping result is... Possibly in Interval.

[0135] The second batch of lighting fixtures ( to (To date, approximately 14,400 hours have been accumulated). At a moderate level, corresponding It is in the higher range, and the estimated value is higher than that of the first batch of lamps.

[0136] The third batch of lighting fixtures ( to (Newly installed) ,correspond This is in complete agreement with expectations and also verifies the effectiveness of the benchmarking.

[0137] Between the three batches The gradient difference in the estimated values ​​is the input driving force for the differential compensation optimization in step S6: aging lamps with low luminous flux maintenance will be assigned positive compensation increments. (Appropriately increase the drive) to compensate for the illuminance difference between it and the newly installed lamps, thereby achieving unified reference commands in group control. Within this framework, road surface illuminance uniformity is achieved across batches. The specific usage of the above estimation results will be detailed in Example 3.

[0138] Example 3

[0139] This embodiment outputs an estimated value for the luminous flux maintenance rate of all luminaires along the entire road segment. Based on this, the specific implementation methods of steps S5 and S6 are explained in detail, namely, how to construct the lamp-road illuminance contribution coefficient matrix. Based on this, the differential compensation dimming increment vector that satisfies the illuminance uniformity constraint is solved. It also explains the linkage triggering mechanism of lifespan warning in compensation solution.

[0140] According to CJJ45, road surface illuminance evaluation points are arranged in a grid pattern along the longitudinal and transverse directions of the road. In the specific scenario of this embodiment (two-way four-lane road, road width approximately 22 meters, and light pole spacing...),... (Meters, light poles are arranged in a crisscross pattern along both sides of the road width), and the evaluation point grid is set up as follows.

[0141] Longitudinal division: Within the section between adjacent light poles on the same side, the light poles are equally divided along the longitudinal direction of the road. In one embodiment, a single evaluation section is taken. (The 17-meter spacing is divided into four equal parts, with each section measuring 4.25 meters). For the layout of 48 light poles along the entire road, the calculated number of coverage sections is approximately 24 (based on the sections between adjacent poles on the same side), totaling approximately 96 longitudinal section locations (example value).

[0142] The road is divided laterally, and on each longitudinal section, it is set along the width of the road. In one embodiment, a set of evaluation points are taken. It covers the entire width from curb to curb (including the edges of the two side lanes and the median strip), with a lateral spacing of approximately 4.4 meters between adjacent evaluation points.

[0143] The above grid setting scheme forms approximately There are 1 road surface evaluation point, and the coordinates of the evaluation point are as follows: ( The data is stored in the road network geometric database with the starting point of the road segment as the origin. It should be noted that the number of evaluation points... The selection of grid density should balance computational accuracy and computational cost. The reasonable range depends on the total length of the road segment and the number of light poles, and the above example values ​​do not constitute a limitation on the scope of protection of this invention.

[0144] matrix The construction requires the following geometric parameters, all of which are read from the road network geometry and light distribution database described in Example 1: The... Geographic coordinates of the light pole (Entered during installation); Installation height (The distance from the road surface to the light-emitting surface of the lamp is uniformly 10 meters in this embodiment, but may vary depending on the batch); Cantilever length (Horizontal extension distance of the lamp post cantilever, approximately 1.5 meters in this embodiment); Lamp optical axis pitch angle (installation elevation / pitch angle, 5° downward horizontally in this embodiment, example value). The above parameters are entered into the database once during installation and remain unchanged thereafter unless the lamp is moved or replaced.

[0145] Matrix elements Indicates the first The evaluation points of the light pole under unit normalized luminous flux and full drive conditions. The theoretical illuminance contribution value generated is measured in lux / lm. This is based on the photometric geometry of luminaires. Calculated using the following formula:

[0146] ;

[0147] The meanings of each parameter and how to obtain them are as follows:

[0148] The angle of incidence; the distance from the light-emitting center of the luminaire to the evaluation point. The angle between the line connecting the points and the road surface normal (vertical line) is determined by the lamp coordinate system. Coordinates of the evaluation point The geometric relationships can be calculated directly:

[0149] ;

[0150] The luminous direction angle is the zenith angle and azimuth angle in the luminaire coordinate system, which are the directions from the light-emitting center to the evaluation point. These angles are calculated by combining the luminaire installation orientation with the geometric coordinate difference.

[0151] The luminous intensity is the value of the luminaire in this luminous direction, obtained by bilinear interpolation from the luminaire's photometric data file (IES or LDT format). The photometric data file is entered into the database along with the luminaire's factory specifications. When the luminaire's photometric data file is unavailable, a standard photometric curve template matching the luminaire's model classification code (according to the photometric type classification in CJJ45 appendix) is used for calculation, and the result is displayed in the matrix. The corresponding column is marked with "Template Estimation", and the refinement update of this column is automatically triggered when the actual light distribution file is supplemented.

[0152] In this embodiment, for each light pole ( ) and each evaluation point ( ),calculate And fill in the matrix to form Illuminance contribution coefficient matrix .

[0153] Physically, the contribution of a distant light pole to the illuminance of the evaluation point is due to... The decay rapidly approaches zero. In this embodiment, for light pole-evaluation point combinations with a horizontal distance exceeding three times the installation height (i.e., exceeding approximately 30 meters, equivalent to approximately 1.8 light pole spacing), its The contribution of the value to illuminance is usually less than 1% of the total illuminance, and can be directly set to zero without introducing substantial calculation error. Utilizing the aforementioned sparsity, the matrix... Storing the matrix locally on the edge controller in a sparse compressed format significantly reduces storage footprint and computational complexity of subsequent matrix-vector multiplications. The calculation is performed offline in one go, and the results are persistently stored. Subsequent batch expansions will perform local incremental updates.

[0154] Given the estimated value of the luminous flux maintenance rate of all luminaires in the network Dimming level of each lamp Under these conditions, evaluation points The predicted illuminance is:

[0155] ;

[0156] in For the first The rated initial luminous flux of the luminaire (taken from the luminaire archive). This is the mapping function from dimming commands to the actual output luminous flux ratio. For PWM dimming, the output luminous flux and the dimming duty cycle are approximately linear, so we take... ( For constant current analog dimming, due to the nonlinear change in LED luminous efficacy with current, take... ,index Usually in The range (example range) can be provided by driver characteristic tests or driver datasheets, and should be set according to the actual driver type during implementation.

[0157] Expressed in vector form, let (matrix No. Multiplying the column by the luminous flux coefficient, the illuminance vector of the entire road surface is:

[0158] ;

[0159] Road surface illuminance uniformity is defined by the minimum-mean ratio:

[0160] ;

[0161] Uniformity threshold According to the CJJ45 standard, the threshold for arterial roads is usually no less than 0.4. The above threshold should be read from the road network geometry database according to road level classification, and not hard-coded in the edge controller, so as to support flexible configuration for different road level scenarios.

[0162] Dimming compensation increment for each lamp The following physical boundary constraints must be met:

[0163] ;

[0164] in The lowest low light limit supported by the driver (typically no less than 10% to maintain normal illumination, example value). This is the maximum dimming limit (usually 100% full drive, i.e.) When a certain lamp... The estimated value is as low as ( When the lifespan determination threshold is reached, the luminaire triggers a lifespan warning and enters the priority compensation level. Relaxed to the absolute upper limit allowed by the drive (i.e. (to maintain the illuminance in the vicinity of the luminaire to the greatest extent physically feasible).

[0165] Regarding the number of light poles Not exceeding the set threshold For road sections or independently controlled sections, the following linear programming relaxation method can be used to solve them accurately.

[0166] Introducing dimming level variable For PWM dimming ( The road surface illuminance is a linear function of the dimming level.

[0167] ;

[0168] Set target average illuminance reference value (Optional: the actual average illuminance without compensation, or the design illuminance value specified in CJJ45), the uniformity constraint is equivalently transformed into:

[0169] ;

[0170] The constraint is about This is a linear inequality. An objective function is established based on the approximate linear relationship between the power consumption of each luminaire and the dimming level (in one embodiment, taking...). ,in (assuming full-drive rated power consumption), the complete optimization problem is then transformed into standard linear programming:

[0171] ;

[0172] The aforementioned LP problem can be solved using the simplex method or interior point method solvers built into the edge controller, with computational delays for... The magnitude of these values ​​is typically within the range of milliseconds to seconds, and does not affect the normal dimming cycle. After the solution is completed, .

[0173] When the number of light poles Exceed When, the following shall be adopted. The iterative greedy algorithm for grouping obtains the suboptimal solution.

[0174] First, initialize all lights by setting their initial compensation increment. Calculate the initial road surface illuminance vector and initial uniformity Then group and sort, according to Sort the lamps in ascending order, and select those with the deepest decay. Lighting fixtures with the lowest (lowest) value will be given priority in the compensation candidate set.

[0175] Iterative compensation involves identifying the contribution coefficient to the current lowest illuminance evaluation point in each iteration. The largest number of light fixtures (i.e., the fixtures most effective at improving "dark areas") are applied with a step size of [missing value]. The positive compensation increment is recalculated. and Simultaneously, it examines whether negative compensation increments can be applied to newly installed luminaires in areas with excessive illuminance (evaluation points with illuminance significantly higher than the average) to save energy. Iterates to... The dimming function will terminate when the dimming limit of each lamp is reached or when it converges.

[0176] Optionally, on more resource-constrained embedded controllers, the above iterative process can be terminated by fixing the maximum number of iterations to ensure that the computation delay is bounded.

[0177] When computing resources are extremely limited, or in scenarios requiring rapid response, a tiered approximation strategy is adopted: [The strategy involves] approximating the entire network... Root light fixture Divided into Each degradation level (in one embodiment) (Values ​​can be 3 to 5, example values) luminaires of the same class share the same compensation increment:

[0178] ;

[0179] in This is the weighted average of the luminous flux maintenance rate of all luminaires in the network. For the first Average value of luminaires within the same grade, progressive compensation coefficient It is used to suppress the phenomenon of inverted illuminance at the boundary of adjacent levels caused by level discretization.

[0180] The physical meaning of this formula is: the improvement ratio of the compensation drive level is equal to the luminous flux loss ratio of that level of luminaire relative to the average level of the entire network, multiplied by the asymptotic factor. To achieve a smooth transition. In one embodiment, The initial value can be 0.8. After the self-calibration verification in step S7, If the target is still not met, the value will be automatically increased gradually in increments of 0.05. Up to version 1.0, the parameters mentioned above are examples and can be adjusted according to the actual scenario during implementation. The hierarchical approximation strategy has slightly lower accuracy compared to the aforementioned LP or greedy algorithms, but the computational cost is only [missing information]. Its scale is suitable for embedded scenarios with strict requirements for computational latency.

[0181] After the solution is completed, the edge controller generates individualized dimming commands for each lamp:

[0182] ;

[0183] Then execute the following distribution process.

[0184] Instruction compliance verification: Before issuance, each instruction undergoes an out-of-bounds check to ensure compliance. For the reason Light fixtures that are too low will require a compensation amount exceeding the dimming limit (i.e., ), actually issued At the same time, the "compensation saturation" status is recorded in the file, and a prompt is sent to the operation and maintenance platform, indicating that the light fixture can no longer fully compensate for its light decay through dimming, and it is recommended to prioritize replacing the light fixture.

[0185] The edge controller will compensate the increments of each lamp. Grouping and packaging commands according to the jurisdiction of the central controller, and sending them to the corresponding central controller as NB-IoT unicast messages; the central controller unpacks the commands into DALI unicast frames according to the DALI address of the light pole, and sends them to each driver respectively. Group control broadcast commands Normal transmission occurs on the DALI bus. Each lighting fixture driver receives the unicast frame and then... When broadcast commands are executed, unicast frames have higher priority than broadcast frames.

[0186] Each solution yields Vector along with the corresponding Input values, calculation timestamps, and predicted uniformity It is cached locally on the edge controller. The cache expiration time covers the next cached cache. When an update (i.e., the next step S4 calculation is completed) or an external trigger (new batch of lights connected, driver fault alarm) occurs, the cache expires and the solution is recalculated. Within the validity period, if the group control baseline command... Changes occur, and the compensation increment for each lamp changes. Keep it unchanged, only individualize the target dimming level Follow Synchronous translation eliminates the need for resolving the problem.

[0187] by Three batches of lighting fixtures along the entire road section After estimation is completed, group control reference dimming command Taking a time period as an example, we can illustrate the calculation effect of step S6.

[0188] Before compensation, the three batches of luminaires were operating at a uniform 70% drive. Due to differences in luminous flux maintenance rates, there were significant gradients in illuminance at various evaluation points on the road surface: the first batch of luminaires covered the northern half of the road. to The average illuminance was lower than that of the area covered by the third batch of lighting fixtures (intersections, etc.). to The difference in road surface illumination between the two sections may exceed 15%, leading to unevenness. It does not meet the 0.4 lower limit requirement specified in CJJ45.

[0189] After compensation, the state is as follows: After the solution in step S6 is completed, the first batch of aging lamps are assigned positive compensation increments. (Adjust the drive speed appropriately to compensate for the luminous flux loss), newly installed luminaires ( to The negative compensation increment was allocated. (The drive was appropriately reduced to save energy and avoid excessive brightness in the area), and the compensation increment for the second batch of luminaires was at a moderate level. After differentiated compensation, the illuminance at each evaluation point tended to be balanced, and the road surface illuminance uniformity improved. Raise to no less than The level.

[0190] It should be noted that the aforementioned compensation direction is not a fixed rule, but rather a natural result of the optimization solution, specifically depending on the spatial distribution pattern of the three batches of luminaires and the illuminance contribution of the road surface evaluation points. Under certain spatial distributions, the optimization solution may also require some decaying luminaires to remain unchanged while only adjusting the drive level of their adjacent new luminaires to achieve the uniformity target at a lower total power consumption cost. This invention does not pre-determine the compensation direction, but rather takes minimizing the total power consumption under uniformity constraints as the unified objective, with the optimization solver automatically determining the optimal or suboptimal compensation strategy.

[0191] Example 4

[0192] Based on the aforementioned embodiments 1 to 3, this embodiment fully explains the adjacent pole mutual calibration procedure in step S7, the cold start calibration procedure for historical in-service lighting fixtures, the system expansion process when a new batch of lighting fixtures is connected, the dual-mode compensation scheduling mechanism, and the complete device implementation of the controller and system described in this invention.

[0193] The self-calibration mechanism in step S7 is based on the following physical fact: within the same group control section, if the luminous flux maintenance rate of all lamps is the same, then under the same group control dimming command... Below is the actual power consumption of the same type of lamp (with the same rated power). Under the same ambient temperature conditions, it should conform to a narrowband distribution with driver hardware differences as the primary source of voltage distribution. When a lamp experiences positive voltage due to aging... During systemic drift, the actual power consumption will deviate from the statistical benchmark of the segment power consumption. The amount of power consumption deviation is related to the positive voltage drift caused by aging. There is a direct physical connection. Step S7 utilizes this set of relative deviations as a coarse-grained observable measure of the aging state, in relation to the output of step S4. The estimated values ​​are corrected, thereby achieving system-level self-calibration closed loop without the need for any external reference sensors.

[0194] The edge controller continuously monitors the following environmental and operating conditions to identify the effective time window that meets the self-calibration execution conditions.

[0195] Includes: Temperature stability conditions: Ambient temperature during the intended calibration period. The change must satisfy (This threshold ensures that power consumption interference caused by differences in the thermal environment of different luminaires is negligible). Ambient temperature data can be estimated from the housing temperature time series of any luminaire driver, or obtained from the nearest weather station data from the upper-level urban lighting management platform; both can be used to determine this condition. Meteorological conditions: No precipitation events during the period (precipitation will change the steady-state power consumption distribution by affecting the heat dissipation of luminaires), which can be identified by abnormal fluctuation patterns in the driver housing temperature time series, or obtained from the meteorological data interface. Steady-state operating conditions: All luminaires in the area must have been running continuously for no less than 30 minutes, and the group control dimming command must be active. It remains unchanged during the period.

[0196] In one embodiment, the time window that meets all the above conditions typically occurs between late night and early morning (when temperatures are stable and the probability of precipitation is low). The edge controller automatically scans and records valid windows daily, and performs the calibration procedure within the valid window. If no valid window appears for several consecutive days, a calibration delay alarm is sent to the operation and maintenance platform, prompting manual intervention to confirm whether the environmental data source is normal. The above parameters are examples and do not constitute a limitation on the scope of protection of this invention.

[0197] Within the effective calibration window, the edge controller applies dimming commands to all lamps within the same group control section (with the centralized controller's jurisdiction as the calibration unit; in this embodiment, each unit contains 8 lamp posts). Real-time power consumption Perform statistical modeling.

[0198] Specifically, a normal distribution is fitted to the power consumption of each lamp within the section. ,in:

[0199] ;

[0200] It should be noted that in phased renovation scenarios, different batches of lighting fixtures may coexist within the same group control zone, and their rated power may vary slightly. Therefore, before establishing a statistical benchmark, the power consumption of each lighting fixture should be normalized to its rated power, i.e., based on... (The ratio of actual power consumption to rated power consumption) replaces Incorporate statistical analysis to eliminate the impact of batch-to-batch variations in rated power. Here... Read from the lighting archive.

[0201] Lighting fixtures within the same section that meet the following conditions are identified as having abnormal power consumption statistics:

[0202] ;

[0203] in The standard deviation threshold for anomaly detection can be set in one embodiment. This corresponds to the normal range at approximately a 95% confidence level (example value). For the identified abnormal luminaires, calculate their luminous flux maintenance correction:

[0204] ;

[0205] in The power consumption-luminous efficacy correlation coefficient reflects the linear proportional relationship between the power consumption deviation and the luminous flux maintenance rate deviation. The data are determined by linear regression from measured data of the same type of lamps at different age stages and stored in the lamp archive. When the calibration data is insufficient, the prior values ​​of literature statistics can be used as the initial values, and then dynamically updated as the system accumulates data.

[0206] Revised The historical estimation sequence is incorporated through an exponentially weighted moving average (EWMA) filter to suppress estimation abrupt changes caused by occasional disturbances.

[0207] ;

[0208] Forgetting factor The selection of [aspect name] balances the degree of preservation of historical estimates with the speed of response to new revisions. When the value is large, the system responds slowly to a single correction, suitable for normal, stable in-service scenarios; when the value is small, the response is fast, suitable for the accelerated aging phase. In one embodiment, The default value can be set to 0.90. In practice, it can be adjusted according to the typical decay rate of the lamp model. The above value is only an example.

[0209] Updated The historical time sequence records are synchronously written into the lighting fixture archive and trigger step S6 to re-solve the compensation increment in the next dimming control cycle, forming a complete closed loop of "acquisition - estimation - self-calibration - compensation".

[0210] When the system of this invention is first deployed on a road section with existing in-service lighting fixtures (without installation timestamps or reference fingerprint records), the following cold start calibration procedure is executed to assign an initial value to the historical in-service lighting fixtures. Estimated value.

[0211] During the initial planned maintenance power outage restart following the system's initial deployment, the luminaire junction temperature has cooled to near ambient temperature due to the power outage, and the luminaire's aging state can be considered frozen during this brief period. Under these conditions, immediately after power-on (before thermal steady-state is established), the rated drive current should be applied. Acquisition of cold-state forward voltage .

[0212] Since the junction temperature is approximately equal to the ambient temperature at this time ( Cold forward voltage Influenced only by its current aging state, it can serve as a pseudo-benchmark containing aging information. Compared with the same model of lamps temperature, Factory initial reference voltage under current conditions (Compared with data from the lamp model data sheet or measured values ​​from newly installed lamps in the same batch), the estimated cold-state aging offset is obtained:

[0213] ;

[0214] Assigned based on a single cold measurement The initial estimate has significant uncertainty, so Bayesian inference is used to introduce prior knowledge to narrow the estimation interval.

[0215] Prior distribution can be queried using a combination of lamp model and cumulative runtime. Among them, the cumulative duration estimate The prior distribution's mean and variance are derived from maintenance records (if any) or lamp replacement logs for the same area, and are retrieved from the LM-80 / TM-21 statistical database for the same model of lamps. The likelihood function is based on the mapping function calibrated in step S4, and is derived from... Construct about likelihood function And consider the uncertainty broadening of measurement noise. The posterior estimate is obtained by Bayes' theorem, which yields the posterior distribution. Take the posterior mean as The initial cold start value is used, and the posterior standard deviation is taken as the initial confidence level reference for the luminaire, assigning it a lower confidence score. (The low confidence flag will be retained until sufficient self-calibration correction data is accumulated in the subsequent step S7 before it can be removed.)

[0216] As the in-service data continues to accumulate, the aforementioned Bayesian posterior initial values ​​will be gradually replaced by direct estimations based on the physical mappings in steps S3 and S4, thus completing the natural exit from the cold start state.

[0217] When a new batch of luminaires is added to the existing group control network, the edge controller automatically executes the following access process to ensure a smooth integration of the old and new batches and timely restoration of illuminance uniformity, without interrupting the normal operation of the existing luminaires.

[0218] Step 1: New Light Fixture Registration and Baseline Fingerprint Collection: After the installation and commissioning of the new batch of light fixtures, the edge controller detects a new light fixture address in the network, triggering the registration process. This involves executing the complete baseline fingerprint registration process for the new light fixture, writing its file into the light fixture database, and setting its initial luminous flux maintenance rate to [value missing]. Confidence score (The status of the new lighting fixtures has been confirmed.)

[0219] The second step is a partial update of the road network geometry database: The geographical coordinates, pole height, cantilever parameters, and light distribution curve data of the new luminaires are entered into the road network geometry database, triggering the matrix. Incremental reconstruction. Specifically, only the affected road surface evaluation points around the location of the new light fixture (within a horizontal distance of no more than 3 times the installation height) are recalculated. Value, for matrix Perform local row and column updates instead of triggering a full matrix recalculation to reduce computational overhead and shorten update latency.

[0220] Step 3: Incremental re-optimization of compensation for affected sections: matrix After the partial update is completed, the compensation incremental optimization in step S6 is re-executed for the road sections within the coverage area of ​​the new luminaire (including the new luminaire itself and several adjacent existing luminaires), generating updated individualized dimming commands, which are then issued in the next dimming control cycle. The dimming commands for luminaires in unaffected sections remain unchanged throughout the process, ensuring their normal operation.

[0221] The fourth step is to include the new luminaire in the self-calibration participant set: After the new luminaire has completed the accumulation of power consumption data for at least one complete and effective calibration window, it is included in the neighboring pole mutual calibration participant set in step S7, and participates in power consumption statistical modeling and anomaly detection together with the existing luminaires to establish its statistical benchmark in the group control network.

[0222] The core principle of the above access process design is that each operation is based on local incremental updates to avoid global recalculation or network-wide shutdown caused by new batches of access, thus ensuring the continuity and real-time nature of system expansion.

[0223] This invention supports two compensation scheduling modes, and the edge controller can automatically switch according to the operating status.

[0224] Periodic Static Compensation Mode: This is the default operating mode. In this mode, the edge controller performs step S4 in batches according to a set period (once per calendar day in one embodiment, configurable from 1 to 7 days). The compensation increment is re-optimized in update and step S6, and all lamps maintain a fixed compensation increment vector within the day. Unchanged. This mode has low computational resource consumption and is suitable for normal operation when the aging of lamps progresses smoothly and the uniformity of illuminance changes slowly.

[0225] Event-triggered dynamic compensation mode: The system automatically switches to dynamic compensation mode and immediately triggers local or global re-estimation and re-optimization when the following events occur: First, the self-calibration module in step S7 detects that the power consumption deviation of a certain lamp exceeds the dynamic alarm threshold, indicating that the lamp may experience sudden acceleration of light decay; Second, the event of a new batch of lamps being connected is triggered; Third, the power consumption deviation of a certain lamp... The estimated value fell to [value] for the first time. Lifetime threshold This triggers a lifespan warning.

[0226] In dynamic compensation mode, the re-optimization calculation after an event is triggered covers the affected local segment, and the new compensation command is issued within the next dimming control cycle after the event occurs to ensure bounded response delay. After the event is processed, the system automatically returns to the periodic static compensation mode to continue operation.

[0227] In some embodiments, a group control illuminance equalization system for LED streetlights of different ages is formed. The system includes: a plurality of LED streetlights, each equipped with a device capable of measuring and reporting positive voltage. Drive current and driver housing temperature It also supports intelligent drivers that receive individualized dimming commands; a hierarchical group control communication network that connects the intelligent drivers of each luminaire with the group control illuminance equalization compensation controller, supporting concurrent transmission of unified group control broadcast commands and individualized compensation unicast commands; a group control illuminance equalization compensation controller; and a road network geometry and light distribution database that stores the geographical coordinates, pole height, cantilever parameters, and light distribution curves of each luminaire, providing a basis for the illuminance contribution matrix. The initial construction and incremental update usage; and the phased transformation batch access management module, which automatically completes the new light fixture baseline fingerprint collection, partial update of the road network database, and matrix when a new batch of lights is added to the group control network. Incremental reconstruction and full-segment compensation vector The system is highly optimized; without adding any new road surface or lamppost photometric sensors, it relies solely on the inherent electrical monitoring data of the lamp driver to estimate the luminous flux decay state of luminaire networks of different ages and compensate for the road surface illuminance uniformity, thereby improving the road surface illuminance uniformity. Meets the design standard threshold Requirements.

[0228] The hierarchical group control communication network adopts the following topology: the bottom layer is a short-range communication link between the intelligent drivers of each light pole and the central controller, supporting DALI protocol, RS-485 bus or ZigBee wireless communication; the upper layer is a wide area network communication between the central controller and the group control illuminance equalization compensation controller, supporting NB-IoT, LoRa or 4G / 5G cellular networks; individualized dimming commands. After being encapsulated, the message is sent to the central controller as a unicast message. The central controller then unpacks the message according to the lamp post address and distributes it to the corresponding lamp driver. The unicast message includes a timeliness verification field to prevent expired compensation instructions from being executed incorrectly.

[0229] The phased renovation batch access management module also triggers adjustments to existing lighting fixtures when a new batch of lighting fixtures is connected. The global reordering of states integrates the old and new batches of lamps into the compensation increment solution in step S6, and drives the lamp life warning and priority compensation module to complete the priority update, so as to ensure the global consistency of the optimization of the uniformity of illuminance in the whole section when the old and new batches coexist.

[0230] In some embodiments, the system or method includes a group-controlled illuminance equalization compensation controller (hereinafter referred to as "edge controller"), which is an industrial-grade embedded controller with local computing and storage capabilities. The controller includes: a driver data acquisition interface module, which periodically acquires the positive voltage of each lamp driver through DALI, RS-485, NB-IoT or PLC carrier communication protocols. Drive current and driver housing temperature The temperature-aging decoupling and light decay estimation module stores the initial voltage reference fingerprint of each lamp. Calculate the forward voltage drift caused by aging based on real-time electrical data. The luminous flux maintenance rate of each lamp is estimated by the calibration mapping function. and its confidence score The illuminance equalization optimization module stores the road surface illuminance contribution coefficient matrix. Receive unified reference dimming command and all lamps Vector, locally solved for differential compensation incremental vectors that satisfy uniformity constraints It does not rely on real-time calculations from an upper-level platform to participate in uniformity closed-loop control; the individualized command issuing module will send individualized dimming commands. The command is sent to the corresponding luminaire driver via unicast, coexisting with the group control broadcast command channel without interference; the self-calibration and full lifecycle file management module executes the neighbor pole mutual calibration self-calibration algorithm, and maintains the installation sequence file, reference fingerprint database, and history of each luminaire. Chronology, History Time-series and confidence score records support batch dynamic expansion and rolling file updates.

[0231] The self-calibration and full lifecycle record management module maintains an independent electrical-aging digital twin record for each luminaire, which includes an installation timestamp. Batch identification, rated initial luminous flux Identification of the obtained junction-shell thermal resistance Temperature coefficient Initial baseline fingerprint ,history Time series and its rate of change ,history Time series and confidence score The files are persistently stored in the controller's local non-volatile memory and are periodically synchronized to the upper-level city lighting management platform, supporting full lifecycle health status traceability.

[0232] The controller also includes a lamp life warning and priority compensation module: when any lamp's lifespan is... Descending to Lifetime determination threshold At that time, an automatic lamp replacement warning is sent to the operation and maintenance platform, and the upper limit of the compensation increment for the lamp is set in the constraint optimization of step S6. The dimming range is relaxed to its maximum allowable value, and the uniformity of illumination on the road section is maintained to the greatest extent physically feasible until the manual replacement of the lights is completed.

[0233] Driver data acquisition interface module; This module is responsible for establishing and maintaining the communication connection between the edge controller and each centralized controller. It receives electrical telemetry data triplets reported by each lighting driver according to a set acquisition cycle via NB-IoT or other wide area network protocols. The module performs compliance checks on the data. Valid data is written to the local time-series database in separate tables according to the light pole number; non-compliant data is recorded in the exception log and triggers corresponding alarms. This module is also responsible for sending compensation dimming command packages to the central controller, monitoring command confirmation receipts, and implementing a resend mechanism for light fixtures that have not received confirmation.

[0234] Temperature-aging decoupling and light decay estimation module; This module is the core algorithm module of the present invention, implementing all calculations in steps S3 and S4. This module persistently stores a reference fingerprint lookup table for each lamp. Thermal resistance identification value Temperature coefficient and calibration mapping parameters , ; Receive in-service electrical data from the driver data acquisition interface module, and sequentially perform junction temperature estimation, temperature-aging decoupling, aging drift extraction, and luminous flux maintenance rate mapping calculation, outputting the data for each luminaire. Estimated value and confidence score ; and execute the mapping parameter regression calibration procedure according to the set period, updating and .

[0235] Illuminance equalization optimization module; this module performs all calculations in steps S5 and S6. This module persistently stores the pavement illuminance contribution coefficient matrix. (Stored in sparse compressed format), receiving the full network output from the temperature-aging decoupling and optical decay estimation module. Vector and group control reference dimming commands from upper layer or local configuration Based on the road segment size, it automatically selects one of the three solution methods mentioned above—LP relaxation, iterative greedy algorithm, or hierarchical approximation—and outputs a differentiated compensation incremental vector. Simultaneously, maintain the compensation increment cache and validity period flag for each lamp, and respond within the validity period. When changes occur, only incremental updates are performed without triggering a full re-optimization, in order to reduce unnecessary computational overhead.

[0236] All optimization calculations for the illuminance equalization optimization module are performed on the local processor of the edge controller, without relying on real-time instructions from the upper-level city lighting management platform for uniformity closed-loop control. This module only reports summary data of the health status of each luminaire to the upper-level platform (including...). The estimated value, confidence score, compensation saturation flag, and uniformity compliance status are not visible to the upper-level platform and do not affect the autonomous operation of the edge controller.

[0237] Individualized instruction issuing module; the individualized instruction issuing module receives the compensation increment vector output by the illuminance equalization optimization module. Calculate individualized dimming commands for each light fixture. Perform cross-boundary verification (to ensure) Afterwards, the data is grouped and encapsulated into unicast command messages according to the jurisdiction of the central controller. These messages are then sent to the corresponding central controller via the NB-IoT link. The central controller then unpacks the messages according to the DALI address of the light pole and sends them to each driver. The unicast command message header includes a target light pole address field and a validity verification field (the validity period is bound to the current dimming control cycle). Expired messages are discarded by the central controller and are not executed. This module also receives the command confirmation acknowledgment from the central controller and transmits it to the driver's data acquisition interface module, forming a closed-loop confirmation of command issuance.

[0238] The self-calibration and full lifecycle record management module runs the adjacent pole mutual calibration self-calibration algorithm, periodically scans the effective calibration window, and performs power consumption statistical analysis. The revised estimate is then fed back to the temperature-aging decoupling and light decay estimation module.

[0239] This module also maintains an electrical-aging digital twin file for each light fixture. The file uses the pole number as the primary key and records the following fields: installation timestamp. (Accuracy to the hour); Batch ID; Rated initial luminous flux The obtained junction-shell thermal resistance was identified. and forward pressure drop temperature coefficient Initial positive voltage reference fingerprint lookup table ;history Time series and its current rate of change ;history Time series and current confidence score Compensation saturation indicator and lifespan warning indicator.

[0240] The aforementioned archive data is persistently stored in the edge controller's local non-volatile memory, supporting complete state recovery after system restart; and is synchronized to the upper-level city lighting management platform at set intervals, supporting maintenance personnel to trace and query the health status of all road sections' lights throughout their entire lifecycle.

[0241] In addition, this module also implements a lifespan warning and priority compensation linkage function. When any lamp... The estimated values ​​were lower than twice in a row. Lifespan determination threshold ( When this happens, the module sends a light replacement warning to the operation and maintenance platform. The warning includes the light pole number and the current... The estimated value, cumulative runtime, and remaining available compensation increment margin; simultaneously, the illuminance equalization optimization module is notified of the upper limit of the compensation increment for this luminaire. Relaxed to the absolute upper limit of the dimming range (i.e.) To the greatest extent physically feasible, maintain the illuminance uniformity of the area adjacent to the luminaire to the required standard until the manual replacement of the luminaire is completed and the new luminaire registration process is triggered.

[0242] The following is an illustration of a typical implementation scenario, using the scenario described in Example 1 (800-meter main road, 48 light poles, and phased renovation in three batches).

[0243] Each LED street light fixture is equipped with an intelligent driver that meets the following requirements: It has a positive voltage. Drive current and driver housing temperature It has local real-time measurement and periodic reporting capabilities; it supports simultaneous reception of broadcast group control dimming commands. With unicast individualized compensation instructions Unicast commands have higher priority than broadcast commands; the DALI protocol or other equivalent single-lamp communication protocols are supported. The intelligent driver does not perform any light attenuation estimation or compensation optimization calculations, but is only responsible for data reporting and command execution to minimize the hardware and software complexity at the luminaire end.

[0244] The communication network adopts a two-layer topology. The bottom-layer communication link connects the intelligent drivers of each light pole to the central controller. In this embodiment, a DALI wired bus is used (each central controller manages 8 light poles), supporting concurrent transmission of broadcast frames (group control commands) and unicast frames (individualized compensation commands). In other embodiments, the bottom-layer link can also use RS-485 bus or ZigBee wireless communication to adapt to the wiring conditions of different installation environments. The choice of the above-mentioned bottom-layer protocol does not constitute a limitation on the scope of protection of this invention. The upper-layer communication link connects the central controller and the edge controller. In this embodiment, an NB-IoT wide area network is used. In other embodiments, LoRa, 4G, or 5G cellular networks can be used instead.

[0245] Personalized dimming commands After being packaged into unicast messages by the edge controller, the messages are sent to the corresponding central controller. The central controller then unpacks the unicast messages according to the DALI address of the light pole and sends them to each driver. The unicast message contains a validity period verification field, and the validity period is strictly bound to the current dimming control cycle. The central controller verifies the validity period before forwarding the message, and messages that have expired are discarded to prevent expired compensation instructions from being mistakenly executed due to communication delays.

[0246] The road network geometry and light distribution database is deployed locally on the edge controller, recording the installation coordinates, pole height, cantilever length, optical axis angle, and light distribution curve files of each luminaire along the entire road segment, providing data for the illuminance contribution matrix. Used for building and incremental updates.

[0247] The database also maintains a light distribution type classification code field for each luminaire. In a typical scenario of this embodiment, the three batches of luminaires have the same nominal model, and their light distribution curve files can share the same file; when the light distribution curves of different batches of luminaires differ slightly due to production batch variations, each batch stores its light distribution file independently. When the measured light distribution data for a luminaire is unavailable, an industry-standard light distribution template (classified according to the light distribution type as specified in Appendix CJJ45) matching its light distribution type classification code is substituted into the matrix. The corresponding column is calculated, and the column is marked as a template estimate in the database; when the measured light distribution file of the luminaire is subsequently entered into the database, the system automatically triggers the matrix. The corresponding columns are refined and updated, and the compensation incremental optimization in step S6 is re-executed to utilize more accurate photometric data.

[0248] The phased renovation batch access management module operates as a sub-function of the self-calibration and full lifecycle file management module in the software. This module monitors lamp address change events in the group control network; when a new lamp address is detected, it automatically initiates the baseline fingerprint filing process described in Example 1; after filing is completed, it sequentially triggers the partial update of the road network geometry database and the matrix... Incremental reconstruction, incremental re-optimization of compensation for affected sections, and instruction update issuance are all performed. New batches of luminaires are included in the self-calibration participant set in step S7, and participate in network-wide mutual calibration after accumulating sufficient calibration window data. The entire access process requires no manual intervention and does not interrupt the normal dimming operation of existing luminaires.

[0249] In addition, when a new batch of lighting fixtures is connected, this module simultaneously triggers the self-calibration and full lifecycle file management modules to process all existing lighting fixtures in the network. The global refresh of the state sorting incorporates both new and old batches of lamps into the compensation increment solution in step S6, ensuring the global consistency of the illuminance uniformity optimization across the entire road segment after the new batch is connected, and avoiding omissions in the uniformity correction of cross-batch connection areas due to local matrix updates.

[0250] This embodiment fully illustrates the physical basis and execution process of the system's self-calibration closed loop, the cold-start Bayesian initialization method for historical in-service lamps, the incremental update mechanism for phased renovation and expansion access, the switching logic of the dual-mode scheduling strategy, and the device implementation methods of the five functional modules of the controller and each component of the complete system described in this invention. Through the above-mentioned technical means, without adding any road surface or lamp post photometric sensors, this invention utilizes only the inherent electrical monitoring capabilities of the lamp driver to achieve non-intrusive estimation of the luminous flux decay state of the phased renovation of LED street light networks of varying ages and to compensate for the uniformity of road illuminance, thereby improving the uniformity of road illuminance. Continuously meet no less than The design standards require that the current group control system be able to effectively maintain the uniformity of illumination in the context of phased urban road reconstruction, which is a real pain point in the industry.

Claims

1. A method for estimating the luminous flux decay state of a network of luminaires of different ages and for compensating for illuminance balance in group control, characterized in that, The method includes the following steps: S1, Electrical reference fingerprinting: When each luminaire is connected to the group control network, the data of that luminaire under several preset driving conditions is collected. The initial positive voltage response under the condition of driving current. With the temperature of the driver housing Constructing an initial positive voltage reference fingerprint for a two-dimensional index and install timestamp Batch identification and rated initial luminous flux Store them together in the lighting fixture archive; S2, In-service electrical data periodic acquisition: Read the current forward voltage from each lamp driver according to the set acquisition period. Drive current and driver housing temperature ; S3, Decoupling junction temperature estimation from temperature-aging component: with Estimate the current junction temperature of the luminaire and extract the forward voltage drift caused solely by aging. : ; in The forward voltage drop temperature coefficient of the LED junction region. For LED junction-to-case thermal resistance, The reference junction temperature used when establishing the baseline data; S4, Non-invasive estimation of luminous flux maintenance rate: The luminous flux maintenance rate of each luminaire is estimated by converting the values ​​through a calibration mapping function. : ； Where the mapping parameters , Obtained by the self-calibration procedure in step S7 or by regression calibration of a subset of lamps of known age; S5, Construction of the Road Illumination Contribution Matrix: Based on the installation coordinates, pole height, and light distribution curve of each luminaire, a luminaire-road illuminance contribution coefficient matrix is ​​constructed. ,element Indicates the first The light pole is the first The theoretical illuminance contribution value of each road surface evaluation point under unit luminous flux conditions. The total number of road surface evaluation points. This represents the total number of light poles. S6, Differentiated compensation dimming increment solution and distribution: constrained by road surface illuminance uniformity Given the constraints of the luminaire dimming range, and with the optimization objective of minimizing total power consumption, we solve for the individualized dimming compensation increment vector for each luminaire. : ； in The unified baseline dimming command issued by the group control system will be solved to generate individualized dimming commands. The signal is sent to the corresponding lighting fixture driver via unicast. S7, Mutual calibration and self-verification of power consumption statistics between adjacent poles: During the effective nighttime period when weather conditions are stable, the power consumption statistics of each lamp in the same group control section are verified under group control commands. Actual power consumption Perform statistical distribution analysis, using the median power consumption of the segment. As a reference benchmark, the power consumption deviation is calculated to exceed the threshold. luminous flux maintenance correction of luminaires The feedback is then used to update the estimated value in step S4, achieving a system-level self-calibration closed loop without an external photometric reference sensor. The corrected value is then updated. Step S6 is used for calculation in the next compensation cycle.

2. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... In step S3, when the lamp manufacturer's datasheet does not provide thermal resistance... At that time, a multi-condition thermal balance identification procedure was used during the initial installation phase to determine... The luminaire was operated to thermal steady state under at least three different power conditions, and the corresponding data were collected. Solve the following overdetermined linear equation system simultaneously using the least squares method with a set of three variables, and simultaneously identify... and : ； Identification results and Stored in the lighting equipment archive for direct retrieval during subsequent service.

3. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... In step S4, the mapping parameters and The calibration adopts a hierarchical regression strategy: for a subset of lights with clearly defined installation timestamps within the group control network, its... The measured sequence is used as the input feature, and the theoretical luminous flux maintenance rate is calculated based on the IESTM-21 extrapolation model at the corresponding cumulative lighting hours. To supervise the annotation process, power-law model parameters were fitted using the weighted least squares method, and parameter consistency constraints were applied to different batches of the same lamp model; for each lamp... Output synchronously calculate confidence score The confidence score is determined by The monotonicity of the time series, the noise level, and its deviation from the mean of the same batch are jointly determined when... When the value falls below the preset confidence threshold, a manual review flag is sent to the operations and maintenance platform.

4. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... In step S5, the illuminance contribution coefficient matrix elements Calculated using the following optical geometric model: ； in For the first The light pole is pointing towards the road surface evaluation point. Luminous intensity in the direction (obtained by interpolation from the light distribution data). For the first The installation height of the light pole, The angle of incidence of the light ray relative to the road surface normal; When measured light distribution data for a luminaire is unavailable, a standard light distribution curve template matching the luminaire's model classification code is used for calculation, and then the matrix is ​​entered into the matrix. The corresponding column is marked as the template estimate, which is automatically triggered when the actual photometric data is supplemented. Local refinement updates.

5. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... In step S6, the solution for the differential compensation dimming increment is constructed as an optimization problem constrained by the uniformity of road illuminance and the dimming range of the luminaire. The actual illuminance at each evaluation point on the road surface is calculated using the following dimming-luminous flux coupling model: ； in This is a mapping function from dimming commands to output luminous flux, and it determines the PWM dimming mode. For constant current analog dimming method ,index Determined by the type of drive; pavement illuminance uniformity is constrained by a minimum-to-mean ratio: ； Number of lampposts When the value is small, the optimization problem is solved using relaxed linear programming. When it is large, use The grouped iterative greedy algorithm obtains the suboptimal solution, and the optimization result is cached in the controller's local storage, with its validity period extending to the next calibration cycle.

6. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... When the number of light poles in the group control section When the set threshold is exceeded or the controller's computing resources are limited, the solution in step S6 replaces the global optimization with the following hierarchical approximation strategy: the luminaires are approximated based on the estimated luminous flux maintenance rate. Divided into recession levels Lighting fixtures of the same grade share the same compensation increment. The compensation increment for each level is calculated using the following formula: ； in This is the weighted average of the luminous flux maintenance rate of all luminaires in the network. For the first Lighting fixtures within the specified grade The average grade This is an asymptotic compensation coefficient used to suppress the illuminance inversion phenomenon caused by discretization at the boundary of adjacent levels.

7. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... In step S7, the mutual calibration of adjacent pole power consumption statistics is specifically performed as follows: based on the change in ambient temperature within a time period. Furthermore, the absence of precipitation events serves as a valid criterion for determining the calibration window. Within the effective window, the power consumption of each lamp in the same group control section is analyzed. Fitting a normal statistical distribution ; for satisfying For abnormal luminaires, the luminous flux maintenance correction is calculated using the following formula: ； in The power consumption-luminous efficacy correlation coefficient was determined statistically from measured data of the same model of lamps; the corrected value is... Through the forgetting factor An exponentially weighted moving average filter is incorporated into the historical estimation sequence to suppress estimation abrupt changes caused by occasional interference.

8. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... For historical in-service luminaires without initial baseline fingerprint records, the method further includes a cold start calibration procedure: during the first maintenance power outage restart phase after the initial system deployment, when the luminaire junction temperature has returned to ambient temperature, the cold-state forward voltage is collected at this time. As a pseudo-benchmark; combined with the statistical prior distribution of the factory luminous flux maintenance rate corresponding to the lamp model, and the known age of lamps in the same area. Statistical characteristics, inferred from Bayesian methods, indicate historically in-service lighting fixtures. An initial posterior estimate is assigned; as subsequent self-calibrated data accumulates, the posterior estimate gradually converges to a value based on... Direct estimates of physical mappings.

9. The method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in claim 1, is characterized in that... When a new batch of lighting fixtures is connected to the group control network, the method also automatically triggers the phased upgrade batch access process: collecting and storing the initial baseline fingerprint of the new batch of lighting fixtures. The matrix corresponding to the road surface evaluation points within the influence range of the new lighting fixtures and adjacent lighting fixtures. Incremental reconstruction of rows and columns is performed to avoid full matrix recalculation; the compensation incremental optimization of step S6 is re-executed and the issued instructions are updated for the road sections affected by the new batch; the new batch of lamps are included in the self-calibration participation set of step S7, and participate in the whole network mutual calibration after at least one complete calibration window, and the normal dimming operation of existing lamps is not interrupted during the entire access process.

10. A method for estimating the luminous flux decay state of a network of luminaires of different ages and compensating for the illuminance balance of group control, as described in any one of claims 1 to 9, characterized in that... The method supports switching between two compensation scheduling modes: in the periodic static compensation mode, batch updates are performed on a daily basis. With compensation increment vector Within the same day, a fixed compensation increment is used; in the event-triggered dynamic compensation mode, when the self-calibration module in step S7 detects that the power consumption deviation of a certain lamp exceeds the dynamic threshold, or when the new batch access event described in step S9 occurs, the local segment is immediately triggered. Revaluation and The optimization process ensures that the new compensation command is issued within the next dimming control cycle after the event occurs.

Images