An electronic rectifier for automatically matching lamp voltage

Abstract

This invention relates to the field of electronic rectifiers and discloses an electronic rectifier for automatically matching lamp voltage, comprising: an electrical characteristic acquisition module; a feature extraction and recognition module connected to the electrical characteristic acquisition module; a drive parameter mapping module connected to the feature extraction and recognition module; a drive parameter optimization and prediction module connected to the output terminals of both the electrical characteristic acquisition module and the drive parameter mapping module; and a power output adjustment module connected to the drive parameter optimization and prediction module. The electrical characteristic acquisition module includes: a transient signal acquisition unit, a steady-state signal acquisition unit, and a sensor interface and conditioning unit. This invention achieves high-precision automatic identification and rapid adaptation of the type of lamp by acquiring its transient electrical characteristics at the initial stage of lamp startup, extracting a unique electrical fingerprint using signal processing methods such as continuous wavelet transform, and then combining this with a neural network for initial parameter mapping.

Description

Technical Field

[0001] This invention relates to the field of electronic rectifier technology, and in particular to an electronic rectifier that automatically matches lamp voltage. Background Technology

[0002] Electronic ballasts are an indispensable component of modern lighting systems, widely used to drive various gas discharge lamps, such as fluorescent lamps, high-intensity discharge lamps (HID lamps), and increasingly popular solid-state lighting sources, such as light-emitting diode (LED) arrays. Their main function is to convert mains power into high-frequency AC or specific DC power suitable for the startup and stable operation of specific lamps. To adapt to the electrical requirements of different lamps, existing electronic ballasts typically integrate control circuitry. This circuitry detects certain electrical parameters after the lamp is connected, such as voltage, current, or power, to attempt to match an appropriate driving strategy. Some designs also consider monitoring the lamp's operating status and adjusting the output according to preset logic to ensure the basic operation of the lamp and the safety of the system.

[0003] However, with the continuous advancement of lighting technology and the increasing variety of lamp types, new challenges have been posed to the performance of electronic ballasts. For example, in terms of the accuracy and breadth of lamp type identification, relying solely on steady-state parameters is insufficient to distinguish lamps with subtle electrical characteristics but significantly different driving requirements. There is also room for improvement in the automatic adaptation capability for newly emerging or non-standard lamps. Regarding the optimization of driving parameters, finding a more ideal balance among multiple dimensions such as luminous efficacy, energy consumption, lamp life, and grid compatibility to fully realize the potential of lamps is a direction the industry continues to explore. Furthermore, how to more promptly detect and proactively compensate for the dynamic changes in lamp performance parameters during long-term use, such as characteristic drift due to aging, to maintain lighting quality and extend the effective operating time of luminaires, is also a problem that designers need to address. Simultaneously, facing rapidly developing lamp technology and constantly evolving energy efficiency standards, the flexibility of the electronic ballast's control strategy and its compatibility with future technologies have become important considerations for evaluating its advancement. Summary of the Invention

[0004] The purpose of this invention is to provide an electronic ballast that automatically matches the voltage of lamp tubes, solving the problems of limited recognition ability, insufficient parameter setting, and inadequate dynamic adaptability of existing electronic ballasts in automatically, accurately, and optimally matching different lamp tubes.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An electronic rectifier that automatically matches lamp voltage includes:

[0007] The electrical characteristic acquisition module is used to collect the electrical signals of the connected lamp tubes and convert them into electrical characteristic data;

[0008] The feature extraction and recognition module, connected to the electrical characteristic acquisition module, is used to extract characteristic information representing the lamp tube from the electrical characteristic data and identify the type of lamp tube;

[0009] The driving parameter mapping module is connected to the feature extraction and recognition module. Based on the feature information and the recognized lamp type, it maps and generates a set of initial driving parameters.

[0010] The drive parameter optimization and prediction module is connected to the output of both the electrical characteristic acquisition module and the drive parameter mapping module. It is used to perform multi-objective optimization based on electrical characteristic data and initial drive parameters, combined with the predicted future state of the lamp, to obtain the final drive parameters.

[0011] The power output regulation module, connected to the drive parameter optimization and prediction module, is used to control the power conversion circuit according to the final drive parameters and output regulated electrical energy to the lamp tube.

[0012] Preferably, the electrical characteristic acquisition module includes:

[0013] The transient signal acquisition unit is used to acquire the instantaneous current and voltage waveforms of the lamp at high speed during the lamp start-up phase;

[0014] The steady-state signal acquisition unit is used to periodically acquire the real-time current and voltage waveforms during the stable operation phase of the lamp.

[0015] The sensor interface and conditioning unit are used to filter, amplify, and convert the acquired waveforms to analog-to-digital data to generate electrical characteristic data.

[0016] Preferably, the feature extraction and recognition module includes:

[0017] The electrical fingerprint extraction unit is used to perform time-frequency analysis on the transient signal during the lamp start-up phase and extract the electrical fingerprint feature vector of the lamp.

[0018] The real-time electrical characteristic calculation unit is used to calculate the effective value, power factor and total harmonic distortion of the lamp based on the real-time current and voltage waveforms during the stable operation phase of the lamp.

[0019] The lamp type identification unit is used to compare the extracted electrical fingerprint feature vector with the pre-stored lamp standard fingerprint database, and combine it with real-time electrical features to identify the type of lamp or mark it as an unknown type.

[0020] Preferably, the driving parameter mapping module includes:

[0021] A combined feature construction unit is used to combine electrical fingerprint feature vectors and real-time electrical features into a comprehensive feature vector.

[0022] The neural network mapping unit, employing a neural network model, receives the synthesized feature vector as input and outputs initial driving parameters.

[0023] Initial drive parameters include initial suggested values ​​for target voltage, switching frequency, and duty cycle;

[0024] The network model management unit is used to store the structure and parameters of the neural network model, and to adjust and update the model based on the system's self-learning results.

[0025] Preferably, the driving parameter optimization and prediction module includes:

[0026] The lamp state prediction unit uses a state estimation algorithm to predict the key performance state of the lamp in the future based on historical and current electrical characteristic data and drive parameters.

[0027] A multi-objective optimization function building unit is used to define a comprehensive optimization objective function that includes multiple performance indicators such as power loss, luminous efficiency, harmonic distortion, and lamp life.

[0028] The iterative optimization solution unit employs a heuristic optimization algorithm. Based on the initial driving parameters, it iteratively searches and determines the final driving parameters by combining the predicted future states of the lamps and the comprehensive optimization objective function.

[0029] Preferably, the power output regulation module includes:

[0030] The PWM signal generation unit is used to generate a pulse width modulation signal for controlling the power switching device based on the final drive parameters.

[0031] The power stage drive and conversion unit includes power switching devices and corresponding drive circuits and power conversion topologies, which perform chopping, inversion and frequency conversion on the input power according to the PWM signal;

[0032] The closed-loop feedback control unit is used to monitor the voltage and current output to the lamp in real time and compare them with the target values ​​in the final drive parameters.

[0033] Preferably, the time-frequency analysis is performed using continuous wavelet transform, and its calculation formula includes:

[0034]

[0035] Among them, i transient Ψ(t) represents the instantaneous current signal during the lamp start-up phase; Ψ(t) is the mother wavelet function; a is the scaling parameter; b is the translation parameter; ψ *(.) represents the complex conjugate of the mother wavelet function; Ψ CWT (a,b) are wavelet transform coefficients used to construct the electrical fingerprint feature vector.

[0036] Preferably, the neural network model is a radial basis function neural network model, which outputs the initial driving parameter P. initial The calculation formulas include:

[0037]

[0038] Among them, F combined For the comprehensive feature vector; N hidden W represents the number of hidden layer nodes. j C is the weight vector from the j-th hidden layer node to the output layer; j σ is the center vector of the j-th hidden layer node; j Let be the width parameter of the j-th hidden layer node.

[0039] Preferably, the heuristic optimization algorithm employs the quantum particle swarm optimization algorithm, and its particle position update formula includes:

[0040]

[0041] Among them, X i (k) represents the position of the i-th particle in the k-th iteration, and represents a set of driving parameters; p i (k) is the individual attractor of the i-th particle; C mbest (k) is the mean of the best historical positions of all particles; α qpso (k) is the contraction-expansion coefficient; u ij (k) is a random number in the interval (0,1).

[0042] Preferably, the prediction of the key performance state of the lamp over a future period employs a Kalman filter algorithm, and its state prediction update formula includes at least the following:

[0043] State prediction in one step:

[0044] State update estimate:

[0045] in, The predicted state value at time k; for The state estimate at time k-1; u(k-1) is the control input, i.e., the driving parameter, at time k-1; A is the state transition matrix; B is the input matrix; y(k) is the state update estimate at time k; K(k) is the Kalman gain; y(k) is the observation at time k, i.e., part of the real-time electrical characteristics; H is the observation matrix.

[0046] In summary, the present invention has at least one of the following beneficial technical effects:

[0047] 1. This invention achieves high-precision automatic identification and rapid adaptation of the type of lamp by acquiring its transient electrical characteristics at the initial stage of lamp startup, extracting unique electrical fingerprints using signal processing methods such as continuous wavelet transform, and then combining this with neural networks for initial parameter mapping. Compared to existing technologies that rely on steady-state parameters or simple threshold judgments, this solution solves the problems of insufficient ability to distinguish lamps with similar characteristics and difficulty in identifying emerging lamp types, significantly improving the "plug-and-play" compatibility of the rectifier.

[0048] 2. This invention employs advanced multi-objective optimization algorithms such as quantum particle swarm optimization to finely optimize the driving parameters initially mapped by the neural network, focusing on multiple performance indicators such as luminous efficacy, energy consumption, harmonics, and lifespan. This strategy enables the rectifier to find a more balanced driving scheme for a specific lamp. Traditional technologies often use fixed parameters or empirical table lookups, making it difficult to effectively and dynamically balance multiple performance objectives for different lamps. This invention overcomes this limitation, contributing to the optimization of the overall performance of the lamp.

[0049] 3. This invention introduces a lamp state prediction mechanism based on Kalman filtering, enabling real-time estimation and prediction of lamp state drift caused by factors such as aging and environmental changes. Based on this prediction result, the system can proactively adjust driving parameters for compensation. Compared with existing technologies that often employ hysteresis compensation or fixed aging curves, this invention solves the problems of untimely response and inaccurate compensation, thereby more effectively maintaining the stability of long-term lamp operation and potentially extending its actual service life.

[0050] 4. This invention employs an intelligent process for determining driving parameters, from feature acquisition and identification mapping to optimized prediction. In particular, it endows the electronic ballast with adaptive and learning capabilities through an updatable neural network model and a potential lamp DNA database expansion mechanism. Compared to existing ballasts with fixed control logic and difficulty in upgrading, this invention better adapts to the continuous development of lamp technology, handles more unknown or new lamps, reduces the risk of frequent ballast obsolescence due to lamp upgrades, and enhances the system's long-term applicability and robustness. Attached Figure Description

[0051] Figure 1 This is a system module architecture diagram of the present invention;

[0052] Figure 2 This is a schematic diagram of the electrical characteristic acquisition module of the present invention;

[0053] Figure 3 This is a schematic diagram of the feature extraction and recognition module of the present invention;

[0054] Figure 4 This is a schematic diagram of the driving parameter mapping module of the present invention;

[0055] Figure 5 This is a schematic diagram of the driving parameter optimization and prediction module of the present invention;

[0056] Figure 6 This is a schematic diagram of the power output regulation module of the present invention. Detailed Implementation

[0057] The following is in conjunction with the appendix Figure 1 - Appendix Figure 6 The present invention will be further described in detail below.

[0058] Please see the appendix Figure 1 - Appendix Figure 6 This invention provides an electronic rectifier that automatically matches lamp voltage, comprising:

[0059] The electrical characteristic acquisition module is used to collect the electrical signals of the connected lamp tubes and convert them into electrical characteristic data;

[0060] Specifically, this embodiment of the invention provides an electronic ballast that automatically matches the voltage of lamp tubes. The electrical characteristic acquisition module in this electronic ballast is used to collect the electrical signals connected to the lamp tubes and convert them into electrical characteristic data usable by subsequent processing units.

[0061] In one exemplary embodiment, the electrical characteristic acquisition module includes a sensor interface and conditioning unit, a transient signal acquisition unit, and a steady-state signal acquisition unit.

[0062] The sensor interface and conditioning unit is used to directly sense the voltage and current signals of the lamp. This unit may include a voltage sensor and a current sensor.

[0063] For example, the voltage sensor may employ a precision resistor divider network combined with an operational amplifier for buffer isolation. The current sensor may preferably be a closed-loop Hall effect current sensor or a current detection circuit configured with a precision sampling resistor.

[0064] The sensor interface and conditioning unit further includes analog signal conditioning circuitry. This conditioning circuitry may include an anti-aliasing filter, such as a second-order Butterworth low-pass filter. Its cutoff frequency f c It is usually set to the sampling frequency f. s A fraction of that, satisfying the Nyquist sampling theorem. For example:

[0065] f c =f s / N;

[0066] Here, N is typically greater than 2, and can be 2.5 for example. The conditioning circuit may also include a signal amplification circuit for amplifying the weak signal output by the sensor to a range suitable for the input of the analog-to-digital converter (ADC). One of the core components of the sensor interface and conditioning unit is the analog-to-digital converter (ADC). This ADC is responsible for converting the conditioned analog voltage and current signals into digital signals.

[0067] The transient signal acquisition unit, whose control logic and sensor interface are connected to the conditioning unit, is used to control the ADC to perform high-speed data acquisition during the initial startup phase when the lamp is detected to be connected to the electronic ballast. The sampling frequency f during this phase is... s Higher, for example, f s ≥100kHz. Acquisition duration T transient Preset to fully capture the transient process of lamp activation, for example, T transient =10ms. During this period, the acquired digitized instantaneous current and / or instantaneous voltage signals are stored. Number of sample points acquired. It can be determined by the following formula:

[0068]

[0069] in, f is the total number of sampling points during the transient acquisition phase; s T is the sampling frequency during the transient acquisition phase; transient This represents the transient acquisition duration. The series of high-frequency sampling data acquired constitutes transient electrical characteristic data, which forms the basis for subsequent electrical fingerprint extraction.

[0070] The steady-state signal acquisition unit, whose control logic is also connected to the sensor interface and conditioning unit, is activated after the lamp completes startup and enters a stable or quasi-stable operating state. It controls the ADC to operate at a relatively low sampling frequency f. s-steady Perform periodic data collection. For example, f s-steady The frequency can be set between 1kHz and 10kHz. This stage primarily collects current and voltage waveform data during the lamp's steady-state operation, typically covering one or more power frequency cycles. This data is used to calculate the lamp's real-time operating parameters, such as RMS voltage and RMS current. This series of collected steady-state sampling data constitutes the steady-state electrical characteristic data.

[0071] During operation, when a lamp is connected to the electronic ballast, the system first determines the lamp's connection by monitoring changes in the electrical state of the lamp interface (such as sudden current changes or voltage build-up). Once lamp connection is confirmed, the transient signal acquisition unit is triggered. It performs high-speed sampling and digitization of the current and / or voltage signals during the initial lamp startup phase via a sensor interface and conditioning unit. The acquired digital sequence current and / or voltage signals are output as transient electrical characteristic data. This transient electrical characteristic data is then transmitted to the subsequent feature extraction and recognition module for extracting the lamp's electrical fingerprint.

[0072] After the lamp startup process is completed, or after initial identification and application of initial drive parameters, the steady-state signal acquisition unit takes over. It periodically samples the current and voltage signals of the lamp during steady-state operation via a sensor interface and conditioning unit. The acquired digital sequence is output as steady-state electrical characteristic data. This steady-state electrical characteristic data is also transmitted to the feature extraction and recognition module for calculating real-time electrical characteristics. Furthermore, this steady-state data may also be used as input for state observation by the drive parameter optimization and prediction module.

[0073] Through the above structure and workflow, the electrical characteristic acquisition module can capture key electrical information of the lamp from startup to stable operation.

[0074] The feature extraction and recognition module, connected to the electrical characteristic acquisition module, is used to extract characteristic information representing the lamp tube from the electrical characteristic data and identify the type of lamp tube;

[0075] Specifically, in this embodiment of the invention, the feature extraction and recognition module has its input terminal electrically connected to the output terminal of the aforementioned electrical characteristic acquisition module. The core function of this module is to extract information characterizing the essential characteristics of the lamp from the acquired electrical characteristic data, and thereby identify the specific type or category of the lamp. In an exemplary embodiment, the feature extraction and recognition module includes an electrical fingerprint extraction unit, a real-time electrical characteristic calculation unit, and a lamp type recognition unit.

[0076] The electrical fingerprint extraction unit receives transient electrical characteristic data output from the electrical characteristic acquisition module, particularly the instantaneous current or voltage signal during the lamp start-up phase. This unit first preprocesses the input transient signal, such as through digital filtering to eliminate noise introduced during the acquisition process or through baseline correction. Subsequently, the unit processes the preprocessed transient signal using time-frequency analysis methods. For example, continuous wavelet transform (CWT) is preferably used for the transient current signal i. transient (t) is analyzed. The calculation formula for continuous wavelet transform is as follows:

[0077]

[0078] Among them, i transient Ψ(t) represents the instantaneous current signal during the lamp start-up phase; Ψ(t) is the mother wavelet function; a is the scaling parameter; b is the translation parameter; ψ * (.) represents the complex conjugate of the mother wavelet function; Ψ CWT (a,b) are wavelet transform coefficients used to construct the electrical fingerprint feature vector.

[0079] After obtaining the wavelet transform coefficients or scaling map, the electrical fingerprint extraction unit extracts a set of predefined features to form an electrical fingerprint feature vector. This feature vector may, for example, include energy distribution across multiple frequency bands, dominant frequency components, dominant frequency phase, and energy distribution standard deviation. For example, feature E... k This represents the total energy within the k-th preset frequency band:

[0080] Among them, band k This represents the k-th scale (frequency) interval. The dominant frequency can be determined by the transient current i. transient (t) is subjected to a Fast Fourier Transform (FFT) to obtain the frequency point with the largest amplitude in its spectrum. The dominant frequency of the final electrical fingerprint feature vector is output.

[0081] The real-time electrical characteristic calculation unit receives steady-state electrical characteristic data output from the electrical characteristic acquisition module. This data typically consists of current and voltage sample values ​​for one or more cycles during stable lamp operation. Based on these sample values, the unit calculates a series of real-time electrical characteristics characterizing the lamp's current operating state. These characteristics may include, but are not limited to, RMS voltage, RMS current, active power, apparent power, power factor, impedance modulus and phase angle, and total harmonic distortion of the current. The RMS voltage and RMS current can be calculated using the following formula (taking voltage as an example):

[0082]

[0083] in, This represents the number of steady-state voltage sampling points within a calculation window; v steady [j] represents the j-th voltage sample value acquired during steady-state operation. The power factor can be calculated from the active power P. active With apparent power V rms ·I rms The ratio is:

[0084]

[0085] in,

[0086] Among them, i steady [j] represents the j-th current sample value acquired during steady-state operation; vsteady [j] represents the j-th voltage sample value acquired during steady-state operation.

[0087] Total Harmonic Distortion (THD) I The calculation is as follows:

[0088]

[0089] Among them, I h Ih is the effective value of the h-th harmonic current; I1 is the effective value of the fundamental current, Hh max The highest harmonic order is considered. These calculated real-time electrical characteristics form a real-time feature vector and are output.

[0090] The lamp type identification unit has its input connected to the output of the electrical fingerprint extraction unit, and exemplarily, it can also be connected to the output of the real-time electrical feature calculation unit. This unit internally stores a standard lamp fingerprint database. This database stores standard electrical fingerprint feature vectors F for various known lamp types. database,j,fingerprint , where j is the lamp type index. When the electrical fingerprint feature vector F of the current lamp is received. fingerprint The unit then compares it with each standard fingerprint in the database. The comparison process can employ affinity calculation methods, such as a similarity metric based on weighted distance. An example affinity measure is... j The calculation formula is as follows:

[0091]

[0092] Among them, D fp F represents the dimension of the electrical fingerprint feature vector. fingerprint,k F represents the k-th component of the currently extracted electrical fingerprint. database,j,fingerprint,k w is the k-th component of the standard fingerprint of the j-th type of light tube in the database; k The preset weight for the k-th feature component reflects its importance in recognition; scale k This is the normalization or scaling factor for the k-th feature component, and can be selected. The lamp type recognition unit selects the one with the highest affinity to the current fingerprint. max The standard fingerprint. If Affinity max If the affinity exceeds a preset recognition threshold, the corresponding lamp type will be output as the recognition result. If all affinity scores are below this threshold, or below a lower "new type candidate" threshold, the current lamp can be marked as an unknown type or a new type candidate. The recognition result (lamp type or unknown label) is output.

[0093] In the overall workflow of this module, firstly, the electrical characteristic acquisition module collects transient and steady-state signals. Then, the electrical fingerprint extraction unit processes the transient signals to generate an electrical fingerprint. Simultaneously or shortly thereafter, the real-time electrical feature calculation unit processes the steady-state signals to generate real-time features. Finally, the lamp type identification unit compares the electrical fingerprint (and optionally incorporates some information from the real-time features as auxiliary judgment) with the database to output the identified lamp type. The lamp type information, electrical fingerprint feature vector, and real-time electrical feature vector output by this module will serve as important inputs to the subsequent drive parameter mapping module.

[0094] The driving parameter mapping module is connected to the feature extraction and recognition module. Based on the feature information and the recognized lamp type, it maps and generates a set of initial driving parameters.

[0095] Specifically, in this embodiment, the input terminal of the driving parameter mapping module is electrically connected to the output terminal of the aforementioned feature extraction and recognition module. Based on the identified lamp feature information, this module quickly generates a set of initial driving parameters suitable for the lamp. These initial driving parameters provide a good starting point for subsequent fine-tuning. In an exemplary implementation, the driving parameter mapping module includes a combined feature construction unit, a neural network mapping unit, and a network model management unit.

[0096] The combined feature construction unit receives the electrical fingerprint feature vector output from the feature extraction and recognition module. This unit can also receive real-time electrical feature vectors output from the feature extraction and recognition module. The function of this combined feature construction unit is to integrate feature information from these different sources into a unified comprehensive feature vector. For example, if the electrical fingerprint feature vector is F... fingerprint The real-time electrical feature vector is F realtime The combined feature vector F is then formed. combined This can be obtained through vector concatenation:

[0097]

[0098] in: F is the electrical fingerprint feature vector of the lamp tube; realtime This represents the real-time electrical feature vector of the lamp; [.] T F represents the transpose operation of a vector; combined This is the combined feature vector formed by concatenation. This combined feature vector is then fed into the neural network mapping unit.

[0099] The neural network mapping unit connects its input to the output of the combined feature construction unit. The core of this unit is a pre-trained or online-updable neural network model. This model establishes a non-linear mapping relationship between the synthesized feature vector and the initial driving parameters. For example, a radial basis function neural network (RBFNN) is preferably used as this mapping model. Upon receiving the synthesized feature vector, the RBFNN calculates and outputs a set of initial driving parameters. Its output calculation formula can be expressed as:

[0100] The neural network model uses a radial basis function neural network model, and its output initial driving parameters P initial The calculation formulas include:

[0101]

[0102] Among them, F combined For the comprehensive feature vector; N hidden W represents the number of hidden layer nodes. j C is the weight vector from the j-th hidden layer node to the output layer; j σ is the center vector of the j-th hidden layer node; j Let be the width parameter of the j-th hidden layer node.

[0103] The network model management unit stores and manages the models used by the neural network mapping unit. This includes storing the structural parameters of the neural network, such as the number of hidden layer nodes in an RBFNN. It also includes storing the model's weight parameters, such as the center vector, width parameter, and output layer weights of an RBFNN. This unit also supports model update mechanisms. For example, in systems with self-learning capabilities, this unit can load updated model parameters when new training data or performance feedback is obtained. This allows for continuous improvement in the mapping accuracy and adaptability of the neural network mapping unit.

[0104] In the overall workflow of the drive parameter mapping module: First, the combined feature construction unit receives and integrates feature information from the previous module to form a comprehensive feature vector. Then, the neural network mapping unit processes this comprehensive feature vector using the neural network model stored in the network model management unit. Finally, the neural network model outputs a set of initial drive parameters for the current lamp and operating condition. This module quickly provides customized initial drive settings for different lamps through nonlinear mapping. This lays a good foundation for the subsequent drive parameter optimization process, helping to improve the response speed and final effect of the overall control strategy. This implementation ensures a rapid and intelligent conversion from lamp characteristics to the initial drive strategy.

[0105] The drive parameter optimization and prediction module is connected to the output of both the electrical characteristic acquisition module and the drive parameter mapping module. It is used to perform multi-objective optimization based on electrical characteristic data and initial drive parameters, combined with the predicted future state of the lamp, to obtain the final drive parameters.

[0106] Specifically, in this embodiment, the input terminal of the drive parameter optimization and prediction module is electrically connected to the output terminal of the aforementioned electrical characteristic acquisition module and the output terminal of the drive parameter mapping module. This module aims to finely optimize the initial drive parameters and incorporate predictions of the future state of the lamp to generate the final drive command applied to the power output stage. In an exemplary implementation, the drive parameter optimization and prediction module includes a lamp state prediction unit, a multi-objective optimization function construction unit, and an iterative optimization solution unit.

[0107] The lamp state prediction unit receives real-time electrical characteristic data output from the electrical characteristic acquisition module. This unit can also receive current or historical application drive parameters as control input. The core function of this unit is to estimate the lamp's internal critical states and predict its short-term future trends based on the observed electrical characteristics and control input. For example, a Kalman filter algorithm can be preferably used to implement state estimation and prediction. The state update and prediction process of the Kalman filter algorithm mainly includes the following recursive steps:

[0108] State-one-step prediction equation: Prediction error covariance matrix: P(k|k-1)=AP(k-1|k-1)A T +Q;

[0109] Kalman gain calculation: K(k)=P(k|k-1)H T [HP(k|k-1)H T +R] -1 ;

[0110] State update estimate: Updated error covariance matrix: P(k|k)=[IK(k)H]P(k|k-1);

[0111] in: This is the state prediction vector at time k based on information from time k-1. is the optimal state estimation vector at time k-1, which may include lamp aging factor, luminous efficacy index, etc.; u(k-1) is the driving parameter vector applied to the lamp at time k-1; A is the state transition matrix, describing how the system state evolves over time; B is the input control matrix, describing the influence of control input on the state; P(k|k-1) is the prior error covariance matrix at time k; P(k-1|k-1) is the posterior error covariance matrix at time k-1; Q is the process noise covariance matrix; K(k) is the Kalman gain at time k; H is the observation matrix, mapping the state vector to the observation space; R is the observation noise covariance matrix; y(k) is the observation vector at time k, which can be composed of real-time electrical characteristic data. Let be the optimal state estimation vector after fusing observation information at time k; I is the identity matrix. By iterating through the above equations, the current state of the lamp can be obtained. The predicted state for several future steps is then output to the iterative optimization solution unit.

[0112] A multi-objective optimization function building block is used to define a comprehensive performance index for evaluating the merits of driving parameters. This comprehensive performance index is typically a weighted combination of multiple interrelated or even conflicting sub-objectives. For example, a comprehensive optimization objective function J... total It can be defined as:

[0113]

[0114] Among them, P drive J is a set of candidate driving parameter vectors; loss The objective is to minimize the power loss of the rectifier under given drive parameters; J efficiency The goal is to maximize the luminous efficiency of the lamp tube; J THD The total harmonic distortion of the input current is to be minimized; J lifetime Based on the current driving parameters and the predicted lamp status The life expectancy metric being assessed aims to be maximized; loss ,w eff ,w THD ,w life These are the weight coefficients for each sub-objective, and their sum can be 1 or set according to priority. The specific form of these sub-objective functions can be established based on the lamp model and rectifier characteristics. The comprehensive optimization objective function J... total It is provided to the iterative optimization solution unit as a fitness function.

[0115] The iterative optimization solution unit receives initial driving parameters from the driving parameter mapping module. It also receives predicted state information from the lamp state prediction unit and a comprehensive optimization objective function from the multi-objective optimization function construction unit. This unit employs a heuristic optimization algorithm to search for the optimal combination of parameters within the feasible region of the driving parameters that optimizes the comprehensive optimization objective function. For example, the Quantum Particle Swarm Optimization (QPSO) algorithm is preferred. The QPSO algorithm explores the solution space by simulating the behavior of particles in a quantum system. Its core particle position update strategy differs from traditional PSO; a simplified form can be expressed as: first, calculate the average optimal position of all particles:

[0116]

[0117] Where, p i,best (iter) represents the optimal position; N pop The size of the particle population.

[0118] Then, the position of each particle is updated as follows:

[0119]

[0120] Among them, X i (k) represents the position of the i-th particle in the k-th iteration, and represents a set of driving parameters; p i (k) is the individual attractor of the i-th particle; C mbest (k) is the mean of the best historical positions of all particles; α qpso (k) is the contraction-expansion coefficient; u ij (k) is a random number within the interval (0,1). In each iteration, the corresponding comprehensive optimization objective function value is calculated based on the new particle position. The individual historical best position and the global historical best position are then updated. The algorithm iterates until a termination condition is met, such as reaching the maximum number of iterations or the objective function value converging. Finally, the driving parameters corresponding to the global historical best position are output as the final driving parameters. These final driving parameters are then transmitted to the power output adjustment module.

[0121] In the overall workflow of the drive parameter optimization and prediction module: First, the lamp state prediction unit continuously estimates and predicts the lamp state. Simultaneously, the multi-objective optimization function construction unit defines the evaluation criteria. The iterative optimization solution unit, starting with the initial parameters output by the drive parameter mapping module, combines the predicted information and optimization objectives, and iteratively searches using algorithms such as QPSO. Finally, it outputs a set of finely optimized and forward-looking drive parameters. This module, through intelligent optimization and state prediction, enables the rectifier to dynamically adjust its output to achieve optimal overall performance under complex constraints. It ensures that the drive strategy not only adapts to the present but also anticipates the future, thereby improving the overall performance and lifespan of the lamp.

[0122] The power output regulation module, connected to the drive parameter optimization and prediction module, is used to control the power conversion circuit according to the final drive parameters, and output regulated electrical energy to the lamp tube.

[0123] Specifically, in this embodiment, the power output regulation module has its input terminal electrically connected to the output terminal of the aforementioned drive parameter optimization and prediction module. The main function of this module is to precisely control the power conversion circuit based on the received final drive parameters, thereby stably outputting regulated electrical energy to the connected lamps. In an exemplary embodiment, the power output regulation module includes a PWM signal generation unit, a power stage drive and conversion unit, and a closed-loop feedback control unit.

[0124] The PWM signal generation unit receives the final drive parameter vector output from the drive parameter optimization and prediction module. This parameter vector typically contains information such as the target output voltage, target switching frequency, target duty cycle, and current limit. Based on these parameters, this unit generates one or more sets of pulse width modulation (PWM) signals. These PWM signals are used to control the on and off of the power switching devices in the subsequent power stage drive and conversion unit. The frequency of the PWM signal is determined by the target switching frequency, and its duty cycle is mainly determined by the target duty cycle, and may be fine-tuned by the closed-loop feedback control unit. The generated PWM signals are output to the power stage drive and conversion unit.

[0125] The power stage drive and conversion unit connects its control input to the output of the PWM signal generation unit. The core of this unit is the power conversion circuit topology and its drive circuit. The power conversion circuit topology can be selected according to application requirements; exemplary examples include half-bridge, full-bridge inverter circuits, or Buck, Boost, and other DC / DC conversion circuits. This unit includes power switching devices such as MOSFETs or IGBTs. It also includes gate drive circuits that provide suitable drive signals to these power switching devices, as well as passive components related to the topology, such as inductors, capacitors, and transformers. When a PWM signal is applied to the gate drive circuit, the power switching devices periodically turn on and off according to the PWM signal's instructions. This converts the input electrical energy (e.g., DC power rectified and filtered from the mains) into electrical energy with specific voltage, current, and frequency suitable for lamp operation. The converted electrical energy is then supplied to the lamp through the unit's output.

[0126] The closed-loop feedback control unit monitors the actual electrical parameters output from the power stage drive and conversion unit to the lamp. For example, it samples the actual output voltage across the lamp and the actual current flowing through it in real time using voltage and current sensors. This unit compares the sampled actual values ​​with target values ​​from the final drive parameter vector to calculate an error signal. For example, if the control target is the output voltage, the error signal is:

[0127] e V (t)=V target (t)-V actual (t);

[0128] Among them, e V (t) represents the voltage error signal; V target (t) represents the target output voltage value; V actual (t) represents the actual measured output voltage value. This error signal is then input to a fast controller, preferably a proportional-integral-derivative (PID) controller. The output of the PID controller is used to adjust the parameters of the PWM signal generation unit, typically fine-tuning the duty cycle DD. The discrete form of the PID control law can be expressed as:

[0129]

[0130] Among them, u PID [k] represents the output adjustment of the PID controller at time k; e[k] represents the error signal at time k; k is the proportional gain; K i K is the integral gain; d This is the differential gain. This closed-loop feedback circuit can quickly respond to output deviations caused by load changes, input voltage fluctuations, and model uncertainties. It ensures that the actual electrical energy received by the lamp is highly consistent with the expected value.

[0131] In the overall workflow of the power output regulation module: First, the PWM signal generation unit generates a reference PWM waveform based on the optimized final drive parameters. Then, the power stage drive and conversion unit converts the input electrical energy according to this PWM waveform and outputs it to the lamp. Simultaneously, the closed-loop feedback control unit continuously monitors the actual output and feeds back the error signal to the PWM signal generation unit through the PID controller. This allows for real-time and rapid compensatory adjustment of the PWM waveform (mainly the duty cycle). Through precise PWM control and rapid closed-loop feedback, this module can provide stable, efficient, and accurate power supply to lamps with different characteristics. This plays a crucial role in ensuring the lighting quality of the lamps, extending their lifespan, and achieving the overall energy-saving goals of the system. This implementation achieves final fine-tuning and stable output of the lamp drive power.

[0132] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An electronic rectifier that automatically matches the lamp voltage, characterized in that, include: The electrical characteristic acquisition module is used to collect the electrical signals of the connected lamp tubes and convert them into electrical characteristic data; The feature extraction and recognition module, connected to the electrical characteristic acquisition module, is used to extract characteristic information representing the lamp tube from the electrical characteristic data and identify the type of lamp tube; The driving parameter mapping module is connected to the feature extraction and recognition module. Based on the feature information and the recognized lamp type, it maps and generates a set of initial driving parameters. The drive parameter optimization and prediction module is connected to the output of both the electrical characteristic acquisition module and the drive parameter mapping module. It is used to perform multi-objective optimization based on electrical characteristic data and initial drive parameters, combined with the predicted future state of the lamp, to obtain the final drive parameters. The power output regulation module, connected to the drive parameter optimization and prediction module, is used to control the power conversion circuit according to the final drive parameters and output regulated electrical energy to the lamp tube; The electrical characteristic acquisition module includes: The transient signal acquisition unit is used to acquire the instantaneous current and voltage waveforms of the lamp at high speed during the lamp start-up phase. The steady-state signal acquisition unit is used to periodically acquire the real-time current and voltage waveforms during the stable operation phase of the lamp. The sensor interface and conditioning unit are used to filter, amplify, and convert the acquired waveforms to analog-to-digital data to generate electrical characteristic data. The feature extraction and recognition module includes: The electrical fingerprint extraction unit is used to perform time-frequency analysis on the transient signal during the lamp start-up phase and extract the electrical fingerprint feature vector of the lamp. The real-time electrical characteristic calculation unit is used to calculate the effective value, power factor and total harmonic distortion of the lamp based on the real-time current and voltage waveforms during the stable operation phase of the lamp. The lamp type identification unit is used to compare the extracted electrical fingerprint feature vector with the pre-stored lamp standard fingerprint database, and combine it with real-time electrical features to identify the type of lamp or mark it as an unknown type.

2. The electronic rectifier for automatically matching lamp voltage according to claim 1, characterized in that, The driving parameter mapping module includes: A combined feature construction unit is used to combine electrical fingerprint feature vectors and real-time electrical features into a comprehensive feature vector. The neural network mapping unit adopts a neural network model, receives the comprehensive feature vector as input, and outputs the initial driving parameters, which include the initial suggested values ​​of the target voltage, switching frequency, and duty cycle. The network model management unit is used to store the structure and parameters of the neural network model, and to adjust and update the model based on the system's self-learning results.

3. The electronic rectifier for automatically matching lamp voltage according to claim 1, characterized in that, The driving parameter optimization and prediction module includes: The lamp state prediction unit uses a state estimation algorithm to predict the key performance state of the lamp in the future based on historical and current electrical characteristic data and drive parameters. A multi-objective optimization function building unit is used to define a comprehensive optimization objective function that includes multiple performance indicators such as power loss, luminous efficiency, harmonic distortion, and lamp life. The iterative optimization solution unit employs a heuristic optimization algorithm. Based on the initial driving parameters, it iteratively searches for and determines the final driving parameters by combining the predicted future states of the lamps and the comprehensive optimization objective function.

4. The electronic rectifier for automatically matching lamp voltage according to claim 1, characterized in that, The power output regulation module includes: The PWM signal generation unit is used to generate a pulse width modulation signal for controlling the power switching device based on the final drive parameters. The power stage drive and conversion unit includes power switching devices and corresponding drive circuits and power conversion topologies, which perform chopping, inversion and frequency conversion on the input power according to the PWM signal; The closed-loop feedback control unit is used to monitor the voltage and current output to the lamp in real time and compare them with the target values ​​in the final drive parameters.

5. An electronic rectifier for automatically matching lamp voltage according to claim 1, characterized in that, The time-frequency analysis is performed using continuous wavelet transform, and its calculation formula includes: ； in, This is the instantaneous current signal during the lamp start-up phase; For the mother wavelet function; For scale parameters; These are translation parameters; The complex conjugate of the mother wavelet function; These are wavelet transform coefficients used to construct the electrical fingerprint feature vector.

6. An electronic rectifier for automatically matching lamp voltage according to claim 2, characterized in that, The neural network model described is a radial basis function neural network model, which outputs initial driving parameters. Calculation formula include: ； in, This is a comprehensive feature vector; This represents the number of hidden layer nodes. For the first The weight vector from each hidden layer node to the output layer; For the first The center vector of each hidden layer node; For the first The width parameter of each hidden layer node.

7. An electronic rectifier for automatically matching lamp voltage according to claim 3, characterized in that, The heuristic optimization algorithm employs the quantum particle swarm optimization algorithm, and its particle position update formula includes: ； in, For the first The particle in the first The position at the next iteration represents a set of driving parameters; For the first Individual attractors of individual particles; This is the average of the best positions in the history of all particles; The contraction-expansion coefficient; It is a random number within the interval (0,1).

8. An electronic rectifier for automatically matching lamp voltage according to claim 3, characterized in that, The Kalman filter algorithm is used to predict the key performance states of the lamp over a future period, and its state prediction update formula includes at least the following: State prediction in one step: ; State update estimate: ; in, for The predicted state value at time; for State estimate at time 1; for The control input at any given time is the drive parameter; This is the state transition matrix; The input matrix; for State update estimate at time step; Kalman gain; for The observed values ​​at a given time represent some real-time electrical characteristics; This is the observation matrix.

Images