LED illumination energy consumption dynamic tracking and energy-saving optimization method based on edge calculation

Abstract

The invention relates to the technical field of lighting energy consumption analysis, in particular to an LED lighting energy consumption dynamic tracking and energy-saving optimization method based on edge calculation, which comprises the following steps of: performing phase consistency calibration and waveform characteristic compression processing on electric parameter information and dimming instruction information at an edge calculation node; outputting the recessive loss observation sequence; on the basis of the loss component sequence, generating a controllability description result representing the response strength relationship of each loss component to the dimming instruction for distinguishing the suppression effect or amplification effect of the dimming instruction to different loss components, and outputting the loss component sequence and the controllability description result; and correcting the energy-saving dimming instruction sequence according to the verification decoupling result, and outputting an optimized energy-saving dimming instruction sequence. According to the invention, the failure problem that the brightness is globally reduced but the loss of a part of loops is increased in a traditional energy-saving strategy is effectively avoided, and the precision, pertinence and engineering operability of the energy-saving strategy are improved.

Description

Technical Field

[0001] This invention relates to the field of lighting energy consumption analysis technology, and in particular to a method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing. Background Technology

[0002] In recent years, LED lighting systems have been widely used in various scenarios such as architectural lighting, municipal roads, and industrial plants due to their high energy efficiency, long lifespan, and good dimming performance. In order to further improve the energy utilization rate of lighting systems, common energy-saving control strategies mainly include time-based control, duty cycle adjustment, and environmental sensing linkage, supplemented by dimming commands issued by a centralized controller.

[0003] However, existing LED lighting energy-saving solutions still have many technical shortcomings. Traditional lighting energy-saving systems mostly rely on active power, current or total energy consumption for macroscopic evaluation, ignoring non-obvious loss factors such as driver efficiency changes, current waveform distortion (harmonic enhancement) and power factor fluctuations caused during dimming. Therefore, they cannot make a precise judgment on whether dimming operation is truly energy-saving.

[0004] Currently, most energy-saving systems lack quantitative modeling of the relationship between dimming behavior and energy consumption. They cannot identify which dimming operations, under specific driver conditions, grid impedance, or harmonic environments, actually amplify overall energy consumption. This leads to a degree of blindness in dimming strategies, and even counter-energy-saving phenomena such as reduced brightness but increased energy loss. In centralized control architectures, dimming commands are often issued periodically from the upper-level platform, lacking real-time perception and response to local operating states. Dimming execution results are also difficult to transmit back to form a closed loop, limiting the dynamic correction and continuous optimization capabilities of energy-saving control. This makes them particularly unsuitable for complex lighting scenarios with real-time changes in illuminance demand or load characteristics. Summary of the Invention

[0005] This invention provides a method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing. It combines edge computing, implicit energy consumption decoupling, dimming effect modeling and closed-loop control to achieve a more accurate, real-time and adaptive lighting energy-saving optimization control strategy.

[0006] A method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing includes the following steps: S1. Implicit Loss Observable Modeling: At the edge computing node corresponding to the target lamp or target circuit, the electrical parameter information and dimming command information of the LED driver input side are collected synchronously. At the edge computing node, phase consistency calibration and waveform feature compression processing are performed on the electrical parameter information and dimming command information to generate an implicit loss observation sequence for characterizing the current waveform distortion characteristics, power factor change characteristics and drive conversion loss characteristics after the dimming command is applied. The implicit loss observation sequence is then output. S2. Loss Source Controllability Decoupling: The implicit loss observation sequence is input into the controllability decoupler in the edge computing node. The controllability decoupler decouples the energy consumption change of the target lamp or target circuit into a loss component sequence that includes at least the drive conversion loss change caused by dimming and the additional loss change on the power supply side caused by current waveform distortion. Based on the loss component sequence, a controllability description result characterizing the strong relationship between the response of each loss component to the dimming command is generated to distinguish the suppression effect or amplification effect of the dimming command on different loss components. The loss component sequence and the controllability description result are output. S3. Constrained Energy-Saving Command Reconstruction and Edge Closed-Loop Verification: Based on the loss component sequence and controllability description results, the edge computing node reconstructs the dimming command to generate an energy-saving dimming command sequence under the premise of meeting lighting demand constraints. The energy-saving dimming command sequence is issued and executed with the goal of reducing loss components with high controllability and sensitivity to additional losses on the power supply side. After execution, the edge computing node continues to collect and update the implicit loss observation sequence, and inputs the updated implicit loss observation sequence back into the controllability decoupler to obtain the verification decoupling result. The energy-saving dimming command sequence is corrected according to the verification decoupling result, and the optimized energy-saving dimming command sequence is output.

[0007] Optionally, S1 includes deploying a voltage sensor and a current sensor on the LED driver input side of the target luminaire or target circuit to collect instantaneous voltage and current waveforms; simultaneously, deploying a signal capture module at the interface between the dimming controller and the LED driver to collect dimming command waveform data; the voltage sensor, current sensor and signal capture module are all synchronously sampled using the clock source of the edge computing node as the time reference; Edge computing nodes receive synchronously sampled voltage waveforms, current waveforms, and dimming command waveforms; perform correlation analysis on the voltage and current waveforms to determine the initial phase difference between the supply voltage and the load current; align the dimming command waveform data with the current waveform data using timestamps; and determine and compensate for the system delay from the dimming command to the current response by analyzing the time difference between the dimming command transition edge and the current waveform response change edge, generating calibrated data with phase and time alignment.

[0008] Optionally, the waveform feature compression processing includes performing a fast Fourier transform on the current waveform data of a complete power frequency cycle or an integer number of power frequency cycles after calibration, extracting the fundamental current amplitude, the 3rd, 5th, and 7th harmonic current amplitudes, and calculating the total harmonic distortion rate; simultaneously, calculating the displacement power factor and the total power factor based on the voltage and current waveform data of the same cycle; calculating the input active power based on the calibrated voltage and current waveforms, and estimating the instantaneous value of the drive conversion loss power based on the current dimming command value and the pre-stored LED driver efficiency-load characteristic model.

[0009] Optionally, S1 further includes arranging the current waveform distortion features, power factor change features, and drive conversion loss features extracted in each calculation cycle in chronological order, and forming a multi-dimensional observation vector together with the corresponding dimming command value; continuously generating and caching multiple consecutive observation vectors at fixed time intervals to form the implicit loss observation sequence and outputting it.

[0010] Optionally, the controllability decoupler is implemented in the edge computing node as an online parameter identifier including a state-space model or a transfer function matrix; the controllability decoupler receives a latent loss observation sequence as input, wherein the dimming command value in the latent loss observation sequence is used as the control input signal of the system, and the electrical parameter characteristics in the latent loss observation sequence are used as the observation output signal of the system.

[0011] Optionally, the controllability decoupler performs a loss component decomposition and decoupling operation based on the input-output relationship identified by the online system, including: a) Separate the observed total input active power change into a drive conversion loss power change component and an ideal load power consumption change component; b) The power variation component of the drive conversion loss is analyzed into a first sub-component that is directly related to the dimming command, and a second sub-component that varies with the degree of distortion of the current waveform. c) Based on the characteristics of current waveform distortion, calculate the additional loss variation caused by harmonic current on the impedance of the line and power supply system.

[0012] Optionally, the first sub-component directly related to the dimming command constitutes the drive conversion loss change sequence caused by the dimming, and the additional loss change caused by the harmonic current constitutes the power supply side additional loss change sequence caused by the current waveform distortion.

[0013] Optionally, the generation of the controllability description result includes: The controllability decoupler analyzes the rate of change and correlation of each loss component sequence under different dimming command values, calculates the sensitivity coefficient or dynamic response gain of each loss component relative to the dimming command; marks loss components with sensitivity coefficients greater than a first threshold as high controllability components, and those less than a second threshold as low controllability components, and records the correspondence between the dimming command's suppression or amplification effect on each loss component in each subdivision interval, forming the controllability description result; the controllability decoupler outputs the loss component sequence and the controllability description result.

[0014] Optionally, the edge computing node determines a basic dimming command value that meets the minimum illuminance or preset brightness requirements of the current lighting scene based on the loss component sequence and controllability description results and according to the lighting demand model; using the basic dimming command value as a benchmark, it calls a constraint optimization algorithm to find a dimming command parameter that minimizes the sum of loss components that are highly controllable and sensitive to additional losses on the power supply side, while maintaining the lighting demand constraints.

[0015] Optionally, during the iterative optimization process, the constrained optimization algorithm uses the controllability description results to predict the impact of dimming command changes on each loss component, thereby generating an energy-saving dimming command sequence within a specified optimization time window, and sending the energy-saving dimming command sequence to the LED driver of the target lamp or target circuit for execution.

[0016] The beneficial effects of this invention are: This invention proposes to integrate voltage and current waveforms with dimming command signals on edge computing nodes to construct a hidden loss observation sequence. This enables the time-series quantitative expression of various non-visible energy consumption behaviors, including drive conversion loss, harmonic additional loss, and power factor fluctuation. It overcomes the problems of coarse energy consumption assessment and inability to distinguish the source of loss in traditional lighting systems, and provides a precise input basis for subsequent decoupling modeling and command optimization.

[0017] This invention establishes a loss component-dimming influence chain by constructing a state-space model and sensitivity mapping relationship between dimming commands and loss responses, and introduces a controllability grading mechanism. When optimizing dimming commands, it prioritizes suppressing loss components with high controllability and high sensitivity, effectively avoiding the failure problem of traditional energy-saving strategies that globally reduce brightness but increase loss in some circuits, thus improving the accuracy, specificity, and engineering operability of energy-saving strategies.

[0018] This invention, by combining lighting demand functions or lighting-dimming mapping table constraints in edge computing nodes, invokes a constraint optimization algorithm to reconstruct the energy-saving instruction sequence online, ensuring that energy-saving operations do not affect lighting quality. After energy-saving execution, dynamic corrections can be made through re-decoupling and deviation comparison. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the optimization method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the decoupling of the controllability of the loss source in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0022] like Figures 1-2 As shown, a method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing includes the following steps: S1. Implicit Loss Observable Modeling: At the edge computing node corresponding to the target lamp or target circuit, the electrical parameter information and dimming command information of the LED driver input side are collected synchronously. At the edge computing node, phase consistency calibration and waveform feature compression processing are performed on the electrical parameter information and dimming command information to generate an implicit loss observation sequence for characterizing the current waveform distortion characteristics, power factor change characteristics and drive conversion loss characteristics after the dimming command is applied. The implicit loss observation sequence is then output.

[0023] S11, On the LED driver input side of the target lamp or target circuit, deploy voltage and current sensors to collect instantaneous waveform data of voltage and current. ; This represents the instantaneous value of the input voltage acquired at time t, i.e., the voltage waveform. This represents the instantaneous value of the input current collected at time t, i.e., the current waveform; Meanwhile, a signal capture module is deployed at the interface between the dimming controller and the LED driver to collect the instantaneous level or PWM waveform data of the dimming command. , This represents the dimming command value at time t, typically a PWM waveform signal used to control the brightness output of LED lights. All three types of sensor modules use a high-precision clock source from the edge computing node as a time reference for synchronous sampling, ensuring a unified timestamp across the data. .

[0024] S12, Phase Consistency Calibration: The edge computing node receives and samples the results of the phase consistency calibration. , , And perform the following processing: S121, During LED driver operation, a phase difference may exist between its input voltage and current waveforms. This phase difference affects the calculation of power quality indicators such as power factor. Therefore, it is necessary to identify and calibrate this phase difference. A cross-correlation analysis method is used to calculate the initial phase difference between the supply voltage and the load current. The cross-correlation analysis method is as follows: In the edge computing nodes, the voltage waveform within the same sampling window is obtained. With current waveform This ensures that both are sampled synchronously using a unified sampling frequency and clock reference. The sampled data... and Mean or amplitude normalization is performed to eliminate the influence of DC offset or amplitude differences on the cross-correlation results; within a certain time window, the cross-correlation function of voltage and current waveforms is calculated. , is defined as: ;in, This represents the time delay for sliding, measured in sample points; it iterates through a delay range. We obtain the cross-correlation sequence. Then, we find the cross-correlation function. The maximum value corresponds to the delay At this point, the voltage waveform is considered to be... Relative to current waveform The optimal alignment position is Based on the sampling rate and power frequency cycle Delay the time : Output the calculated phase difference .

[0025] S122, in practical applications, after the dimming controller sends a dimming signal, the LED driver does not immediately change the output current; there is a system response delay. If this delay is not calibrated, the dimming command and current change will not correspond, rendering subsequent analysis invalid. The PWM waveform dimming signal exhibits a transition edge, typically an edge transitioning from a low level to a high level; the time is denoted as... Current waveform response, such as a sudden increase or decrease in current occurring at a certain point in time, is denoted as... Based on dimming command waveform With current waveform Timestamp alignment, analysis of dimming signal transition edges With the point of change of current response The time difference between them determines the system response delay. : ;Will For alignment and Generate phase-time aligned calibration data; among which, This indicates the timing of the dimming command's rising edge, such as the rising edge of a PWM signal. This indicates the change in the current waveform response, such as the point in time when a sudden change in current amplitude occurs.

[0026] S13, Waveform Feature Compression Processing: To ensure the physical meaning of the spectrum analysis results, this invention requires sampling the voltage and current waveforms within one or more complete power frequency cycles to form periodic and complete waveform data, that is, sampling within one complete power frequency cycle. Or the current waveform within an integer number of power frequency cycles A Fast Fourier Transform (FFT) is performed to transform the waveform to the frequency domain, identifying its spectral components, particularly the fundamental frequency (50Hz) and lower-order harmonics (150Hz, 250Hz, 350Hz). The harmonic amplitudes are extracted to determine waveform distortion. The result is: Fundamental current amplitude ; Amplitude of each harmonic current ; Total harmonic distortion: The higher the THD value, the more non-fundamental components there are in the current waveform, that is, the more severe the waveform distortion. It is a key indicator for judging the distortion of LED current waveform.

[0027] in, Indicates the fundamental frequency (50Hz) amplitude. These represent the amplitudes of the 3rd harmonic (150Hz), 5th harmonic (250Hz), and 7th harmonic (350Hz), respectively. The 3rd, 5th, and 7th harmonics were chosen because they are the odd-order low-frequency harmonics that have the most significant impact on the power quality of lighting systems, and are particularly noticeable under PWM dimming or nonlinear drive.

[0028] Calculate the following power factor parameters using synchronized voltage and current waveform data: Displacement power factor: It only reflects the phase difference between the fundamental voltage and current waves, without considering the influence of harmonics. Total power factor: Considering the average effect of all waveform components, including harmonic effects, it is a true indicator for evaluating power quality. Indicates the effective value of the voltage. Indicates the effective value of the current; The input active power is: This represents the effective power of the input voltage and current within one power frequency cycle; based on the current dimming command value. Compared with the preset LED driver efficiency-load characteristic model Estimate the drive conversion loss power: ;in, Indicates the power loss of the driver conversion. This is the instantaneous value of the input voltage. This is the instantaneous value of the input current. This indicates a dimming control signal. The voltage and current phase difference Indicates the input active power. Let D(t) be the LED driver efficiency corresponding to the dimming command value. To estimate the driver conversion losses, It is the power frequency cycle. It is a nonlinear function that varies with the dimming duty cycle or load ratio. It is modeled as: ;in, This indicates the dimming command value (duty cycle). This represents the maximum working efficiency, taken as 0.9-0.95. These are all empirical fitting coefficients, reflecting the efficiency decline trend when the load deviates from the rated value. The fitting steps are as follows: In the experimental environment, the dimming duty cycle of the LED driver was changed. Record each Corresponding input power and output power Calculate the experimental efficiency: ; Constructing the fitting objective function: This involves adapting the pre-defined nonlinear model expression. As the fitting function, Curve fitting is performed on the observed data.

[0029] Construct an objective function, using the sum of squared errors across all experimental points as the fitting target: The optimal solution is obtained by using the least squares algorithm. , Compare the fitted model curve with the actual data points to evaluate the fitting error. If the error is large, the model form can be adjusted appropriately, such as by adding higher-order terms or correcting the weights, and then refitting.

[0030] S14, Generate the latent loss observation sequence: Combine the following features extracted from each power grid frequency cycle or sliding window into a multi-dimensional observation vector. : ; Generated in chronological order: This is the latent loss observation sequence, which will serve as the input basis for subsequent controllability decoupling and energy-saving command reconfiguration.

[0031] S2. Loss Source Controllability Decoupling: The implicit loss observation sequence is input into the controllability decoupler within the edge computing node. The controllability decoupler decouples the energy consumption change of the target lamp or target circuit into a loss component sequence that includes at least the drive conversion loss change caused by dimming and the additional power supply loss change caused by current waveform distortion. Based on the loss component sequence, a controllability description result characterizing the strength of each loss component's response to the dimming command is generated to distinguish the suppression or amplification effect of the dimming command on different loss components. The loss component sequence and the controllability description result are then output.

[0032] S21, Decoupling Model Construction and Input: Inside the edge computing node, the controllability decoupler is implemented in the form of a state-space model or a transfer function matrix model, and it operates as an online parameter identifier.

[0033] The input is the latent loss observation sequence from S1: In the model, the dimming command value D(t) at each moment is regarded as the control input signal, and the current waveform characteristics, power factor, loss estimate, etc. are regarded as the observation output signal.

[0034] S22, Loss component decomposition: S221, Decoupling of total input power variation: During each sampling period, the voltage waveform is adjusted accordingly. With current waveform The product integral is used to calculate the input active power. Between two consecutive cycles, the power changed, denoted as This total power change is not entirely caused by LED light emission; it also includes losses due to the internal switching process of the driver. Therefore, it needs to be decoupled into two parts: One part is the power consumption variation component of the ideal load. This refers to the change in lamp load power caused only by dimming, assuming the LED driver efficiency is constant and the waveform is undistorted. The other part is the component of power variation in drive conversion loss. This refers to the additional loss fluctuations caused by factors such as changes in driver efficiency and circuit nonlinearity.

[0035] By calling the LED driver's preset efficiency model, i.e. By calculating the power output under ideal conditions and then subtracting it from the actual input power change, the two components can be separated.

[0036] Specifically, it means the change in input active power at the current moment relative to the previous moment. Represented as: ;in, This represents the component of power consumption variation under an ideal load. This represents the component of the driving conversion loss power variation.

[0037] S222, Internal decoupling of drive losses: The change in drive conversion losses is itself caused by two reasons, therefore... Further decomposed into two sub-components: ;in, This represents the change in drive loss directly related to the dimming command value, i.e., the first sub-component. This represents the loss change caused by waveform distortion, i.e., the second sub-component. The decomposition of these two sub-components can be obtained using the power factor (PF) and total harmonic distortion (THD) indices, along with online identification. The logical relationship between these indices is as follows: 1. A decrease in the total power factor (PF) can be due to either an increase in phase difference or an increase in harmonics. However, when dimming is constant, fluctuations in PF better reflect the efficiency fluctuations of the driver itself, such as a decrease in the efficiency of the control circuit. Therefore, PF is mainly used to estimate the loss components caused by dimming. .

[0038] 2. Total Harmonic Distortion (THD) reflects the proportion of non-fundamental components in the current waveform. The stronger the harmonics, the worse the stability of the driver's magnetic components, rectifier circuit, and current loop, and the greater the losses. Therefore, THD is mainly used to estimate the additional losses caused by harmonics. .

[0039] The online identification process is as follows: 1. In the edge computing nodes, a linear weighted model is used for online fitting, in the following form: ; These are their respective weighting coefficients; Or more explicitly, it can be written as two independent mappings: ; ; in, As a linear function, the model is continuously updated by collecting sample pairs (PF, THD, driving loss). After a sufficiently large sampling window, the online model can converge stably, forming a set of response decoupling rules with local adaptability. The model parameters or coefficients can be obtained through the sliding time window statistical method: retaining data from nearly N observation points, updating the weights in real time according to the minimum mean square error, and finally, in each sampling period, the decoupler outputs two sub-components.

[0040] S223, Harmonic Additional Loss Estimation: Harmonic Current in Line Resistance and Power Supply Impedance The additional losses generated are: ;in, This represents the additional loss of harmonic current on the power supply side impedance. Let be the amplitude of the kth harmonic current. Let be the real part of the equivalent impedance on the power supply side, and represent the active power loss. For the equivalent impedance of the power supply side, each harmonic current component is in Power losses proportional to the square of the current will occur in all of them. By summing the losses corresponding to the 3rd, 5th, and 7th harmonics, we can obtain the additional power supply losses caused by harmonics.

[0041] S224, after the above decoupling, the results obtained in each consecutive sampling period The data is stored in two sequences in chronological order, resulting in the following two loss sequences: 1. Sequence of changes in drive conversion loss caused by dimming: ; 2. Sequence of additional losses on the power supply side caused by current waveform distortion: .

[0042] S23, Controllability Description Result Generation: The core objective is to analyze the impact of dimming commands on various loss components to determine which losses have strong controllability, i.e., high controllability, thereby providing a basis for strategic prioritization in the construction of energy-saving commands. This includes analysis of each loss component. With dimming command The response relationship between them was analyzed to obtain their sensitivity coefficient. or dynamic response gain : ; Let be the sensitivity coefficient of the j-th loss component relative to the dimming command, representing the j-th loss component. The sensitivity to dimming command D refers to how quickly the loss component changes when the dimming command changes. A higher sensitivity indicates a stronger response of the loss component to dimming operations. The dynamic response gain of the j-th loss component is used to measure the transition of the dimming command, indicating the change in the dimming command from the current... Change in the next moment When the j-th loss component changes, assess whether the loss actually fluctuates with dimming. Positive gain indicates amplification, and negative gain indicates suppression. Indicates the sampling time interval. include .

[0043] Classification is based on preset thresholds: like If so, then this component is a highly controllable component. like , then it is a low controllability component; Between these two levels, the level of controllability is moderate. The sensitivity threshold representing the controllability classification is set. This is the upper quantile value, the 80th percentile. The lower quantile and 20th percentile represent the relative strength distribution.

[0044] At the same time, according to The intervals are divided into [0,0.3], [0.3,0.6], and [0.6,1.0], respectively. The average response trend of each loss component in each interval is calculated. The mean value within a certain interval is used; if the mean value is negative, it is judged as a suppression effect; if the mean value is positive, it is judged as an amplification effect. This results in a dimming interval response effect table, indicating the response direction of each loss in each dimming interval.

[0045] The output is structured as follows: 1. Sensitivity coefficient list: ; 2. List of controllability level labels (high, medium, low); 3. Dimming zone response effect table (suppression / amplification).

[0046] S24, Finally, the decoupler outputs the following two structures: Loss component sequence: ; Controllability description results: The energy-saving dimming command, passed as input to S3, is reconstructed to guide it in prioritizing the suppression of components with high controllability and significant impact on total loss. S3. Constrained Energy-Saving Command Reconstruction and Edge Closed-Loop Verification: Based on the loss component sequence and controllability description results, the edge computing node reconstructs the dimming command to generate an energy-saving dimming command sequence under the premise of meeting lighting demand constraints. The energy-saving dimming command sequence is issued and executed with the goal of reducing loss components with high controllability and sensitivity to additional losses on the power supply side. After execution, the edge computing node continues to collect and update the implicit loss observation sequence, and inputs the updated implicit loss observation sequence back into the controllability decoupler to obtain the verification decoupling result. The energy-saving dimming command sequence is corrected according to the verification decoupling result, and the optimized energy-saving dimming command sequence is output.

[0047] S31, the edge computing node reconfigures the energy-saving dimming command based on the following information: Input data: Loss component sequence: ; Controllability description results: ; Lighting demand mapping function: , indicates a dimming command Compared with actual illuminance or brightness The correspondence.

[0048] S311, Lighting Demand Constraint Modeling: Assume the minimum brightness requirement for the current lighting scene is... Then the dimming command must satisfy: It can be based on a preset brightness model. Obtain the minimum illuminance required at present. Basic dimming value , This represents the minimum lighting intensity required for the current scene. The preset brightness model uses a non-linear model: ;in Simulates the nonlinear perception of brightness changes by the human eye. To control the gain coefficient that affects brightness as the dimming value changes, the slope of the brightness curve is determined. The baseline offset for brightness represents the residual brightness when the dimming value is 0. Both are obtained by fitting actual photometric data. Existing mapping models can also be used.

[0049] S312, Construct the objective function and perform constrained optimization: Use the weighted sum of components with high controllability and sensitivity to additional losses as the optimization objective function. : ;in, It is a set of loss components with high controllability, namely the aforementioned The set of loss components, Representation and sensitivity coefficient Proportional weights , Represents a set The Middle The sensitivity coefficient of each loss component, that is, the response strength of that component to a dimming command. It is the dimming sensitivity of each highly controllable loss component, used for weighting. Normalization The dimming command is The corresponding loss estimate for a given dimming command value. The loss component under this command can be estimated using the following formula: Ultimately, the lighting constraints are met. Under the premise of finding the optimal dimming command : ;by Based on this, generate a line of length. Energy-saving dimming command sequence: .

[0050] S32, Execution and Data Update: [This will...] The command is sent to the LED driver of the target luminaire and execution begins in the target luminaire or circuit. The LED driver receives... Dimming instructions at each point in time Then, it will adjust the output current or PWM duty cycle according to the instructions, thereby controlling the brightness output of the LED lamps and achieving the goal of energy saving.

[0051] Based on each instruction value, the lighting fixture behaves as follows within a certain time range: Gradually decrease or increase the brightness; Maintain a minimum brightness level that meets lighting requirements; Minimize driver switching losses and additional losses on the power supply side.

[0052] While executing the sequence, and afterward, the edge computing node continues to execute the synchronous acquisition, calibration, compression and generation process described in step S1, thereby acquiring and updating the latent loss observation sequence. The updated sequence reflects the actual system state after the energy-saving command is executed.

[0053] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0054] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing, characterized in that, Includes the following steps: S1. At the edge computing node corresponding to the target lamp or target circuit, synchronously collect the electrical parameter information and dimming command information of the LED driver input side, and perform phase consistency calibration and waveform feature compression processing on the electrical parameter information and dimming command information at the edge computing node to generate a latent loss observation sequence for characterizing the current waveform distortion characteristics, power factor change characteristics and drive conversion loss characteristics after the dimming command is applied, and output the latent loss observation sequence. S2. Input the latent loss observation sequence into the controllability decoupler in the edge computing node. The controllability decoupler decouples the energy consumption change of the target lamp or target circuit into a loss component sequence that includes at least the drive conversion loss change caused by dimming and the additional loss change on the power supply side caused by current waveform distortion. Based on the loss component sequence, it generates a controllability description result that characterizes the strong relationship between the response of each loss component to the dimming command. This result is used to distinguish the suppression effect or amplification effect of the dimming command on different loss components. The loss component sequence and the controllability description result are then output. S3. Based on the loss component sequence and controllability description results, the edge computing node reconstructs the dimming command to generate an energy-saving dimming command sequence under the premise of meeting lighting demand constraints. The energy-saving dimming command sequence is issued and executed with the goal of reducing loss components with high controllability and sensitivity to additional losses on the power supply side. After execution, the edge computing node continues to collect and update the implicit loss observation sequence, and inputs the updated implicit loss observation sequence into the controllability decoupler again to obtain the verification decoupling result. The energy-saving dimming command sequence is corrected according to the verification decoupling result, and the optimized energy-saving dimming command sequence is output.

2. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 1, characterized in that, S1 includes deploying a voltage sensor and a current sensor on the LED driver input side of the target luminaire or target circuit to collect instantaneous voltage and current waveforms; simultaneously, a signal capture module is deployed at the interface between the dimming controller and the LED driver to collect dimming command waveform data; the voltage sensor, current sensor and signal capture module are all synchronously sampled using the clock source of the edge computing node as the time reference. Edge computing nodes receive synchronously sampled voltage waveforms, current waveforms, and dimming command waveform data; Correlation analysis is performed on the voltage and current waveforms to determine the initial phase difference between the supply voltage and the load current; The dimming command waveform data and the current waveform data are timestamped and aligned. By analyzing the time difference between the dimming command transition edge and the current waveform response change edge, the system delay from the dimming command to the current response is determined and compensated, generating calibrated data with phase and time alignment.

3. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 2, characterized in that, The waveform feature compression processing includes performing a Fast Fourier Transform on the current waveform data of a complete power frequency cycle or an integer number of power frequency cycles after calibration, extracting the fundamental current amplitude, the 3rd, 5th, and 7th harmonic current amplitudes, and calculating the total harmonic distortion rate; simultaneously, calculating the displacement power factor and the total power factor based on the voltage and current waveform data of the same cycle; calculating the input active power based on the calibrated voltage and current waveforms, and estimating the instantaneous value of the drive conversion loss power based on the current dimming command value and the pre-stored LED driver efficiency-load characteristic model.

4. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 3, characterized in that, The S1 further includes arranging the current waveform distortion features, power factor change features, and drive conversion loss features extracted in each calculation cycle in chronological order, and combining them with the corresponding dimming command value to form a multi-dimensional observation vector; continuously generating and caching multiple consecutive observation vectors at fixed time intervals to form the implicit loss observation sequence and outputting it.

5. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 1, characterized in that, The controllability decoupler is implemented in the edge computing node as an online parameter identifier including a state-space model or a transfer function matrix; the controllability decoupler receives a latent loss observation sequence as input, wherein the dimming command value in the latent loss observation sequence is used as the control input signal of the system, and the electrical parameter characteristics in the latent loss observation sequence are used as the observation output signal of the system.

6. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 5, characterized in that, The controllability decoupler performs loss component decomposition and decoupling operations based on the input-output relationship identified by the online system, including: a) Separate the observed total input active power change into a drive conversion loss power change component and an ideal load power consumption change component; b) The power variation component of the drive conversion loss is analyzed into a first sub-component that is directly related to the dimming command, and a second sub-component that varies with the degree of distortion of the current waveform. c) Based on the characteristics of current waveform distortion, calculate the additional loss variation caused by harmonic current on the impedance of the line and power supply system.

7. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 6, characterized in that, The first sub-component directly related to the dimming command constitutes the drive conversion loss change sequence caused by the dimming, and the additional loss change caused by the harmonic current constitutes the power supply side additional loss change sequence caused by the current waveform distortion.

8. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 7, characterized in that, The generation of the controllability description result includes: The controllability decoupler analyzes the rate of change and correlation of each loss component sequence under different dimming command values, calculates the sensitivity coefficient or dynamic response gain of each loss component relative to the dimming command; marks loss components with sensitivity coefficients greater than a first threshold as high controllability components, and those less than a second threshold as low controllability components, and records the correspondence between the dimming command's suppression or amplification effect on each loss component in each subdivision interval, forming the controllability description result; the controllability decoupler outputs the loss component sequence and the controllability description result.

9. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 1, characterized in that, The edge computing node determines the basic dimming command value that meets the minimum illuminance or preset brightness requirements of the current lighting scene based on the loss component sequence and controllability description results and according to the lighting demand model. Based on the basic dimming command value, a constraint optimization algorithm is invoked to find the dimming command parameter that minimizes the sum of loss components that are highly controllable and sensitive to additional losses on the power supply side, while maintaining the lighting demand constraint.

10. The method for dynamic tracking and energy-saving optimization of LED lighting energy consumption based on edge computing according to claim 9, characterized in that, During the iterative optimization process, the constrained optimization algorithm uses the controllability description results to predict the impact of dimming command changes on each loss component, thereby generating an energy-saving dimming command sequence within a specified optimization time window, and sending the energy-saving dimming command sequence to the LED driver of the target lamp or target circuit for execution.

Images