A LED crystal screen with high transmittance and control method

Abstract

The present invention provides an LED crystal truss screen with high transparency and a control method. The method comprises: collecting multi-source sensor data, establishing a nonlinear state space model; using the model to construct a Kaufman nonlinear controller; using a Gauss-Bernoulli mixed model for state estimation; performing fault analysis based on a Pandora time fault tree; and realizing adaptive display control. The present invention improves the display effect and reliability of the LED crystal truss screen by combining Kaufman nonlinear control with dynamic fault diagnosis.

Description

Technical Field

[0001] The present invention belongs to the field of display technology and artificial intelligence, and particularly relates to an LED crystal truss screen with high transmittance and a control method thereof. Background Art

[0002] LED crystal truss screen is a new type of large-size display device that can achieve high transparency display. However, the current LED crystal truss screen control system has the following main problems:

[0003] 1. Existing LED display control methods mainly rely on simple PWM dimming and constant current drive. This control method is difficult to adapt to the nonlinear characteristics of LED devices and environmental interference, and easily leads to unstable display effects. In practical applications, the photoelectric characteristics, temperature characteristics, and attenuation characteristics of LED devices all show obvious nonlinearity, which makes it difficult for traditional linear control methods to achieve ideal control effects.

[0004] 2. Existing technologies lack accurate estimation of system status. LED crystal truss screen is a complex multivariable system, and its internal state (such as LED light intensity, temperature distribution, etc.) is difficult to measure directly. Existing technologies mainly rely on single-point measurement and simple threshold judgment, and cannot make a comprehensive and accurate assessment of the system status, which affects the stability of the display effect.

[0005] 3. Conventional fault diagnosis methods are mainly based on fixed threshold judgments and lack the ability to analyze dynamic faults. This method cannot effectively predict and prevent system failures, resulting in increased maintenance costs. Especially in high-transmittance LED crystal truss screens, the reliability requirements of devices and drive circuits are higher, and existing fault diagnosis methods are difficult to meet the requirements. Summary of the invention

[0006] The purpose of the present invention is to provide a control method and device for an LED crystal truss screen to solve the above technical problems.

[0007] An embodiment of the present invention provides a control method for an LED crystal truss screen, comprising:

[0008] Collecting multi-source sensor data of the LED crystal truss screen, wherein the multi-source sensor data includes LED light intensity data, temperature data, current data and ambient light data;

[0009] Based on the multi-source sensor data, a nonlinear state space model of the LED crystal truss screen is established;

[0010] Using the nonlinear state space model, a Kaufman nonlinear controller is constructed to realize the display control of the LED crystal truss screen;

[0011] Based on the multi-source sensor data, a Gaussian Bernoulli mixture model is used to perform system state estimation to obtain a state estimation result;

[0012] Based on the state estimation result, using Pandora time fault tree to perform dynamic fault analysis and generate fault prediction information;

[0013] When the fault prediction information is received, adaptive display control of the LED crystal truss screen is implemented according to the display control result of the LED crystal truss screen and the fault prediction information.

[0014] Optionally, the establishing of a nonlinear state space model of the LED crystal truss screen includes:

[0015] Establish nonlinear characteristic models of LED devices, including: photoelectric characteristic model describing the current-light intensity relationship, temperature characteristic model describing the effect of temperature on luminous efficiency, and attenuation characteristic model describing the life attenuation law of LED devices;

[0016] Based on the nonlinear characteristic model of the LED device, a state space expression is constructed, and the state space expression includes: establishing a state equation with LED light intensity, temperature, and drive current as state variables, establishing a control equation with PWM duty cycle and drive current parameters as control inputs, and establishing an output equation with actual display effect parameters as system output.

[0017] Optionally, constructing a Kaufman nonlinear controller includes:

[0018] Constructing a Kaufman operator matrix based on the state equation and output equation of the nonlinear state-space model;

[0019] Using the Kaufman operator matrix to lift the nonlinear state space model to a high-dimensional linear space;

[0020] A state feedback controller is designed in the high-dimensional linear space, including constructing a feedback gain matrix and a state observer.

[0021] Optionally, designing a state feedback controller in a high-dimensional linear space includes:

[0022] Designing the elements of the feedback gain matrix by taking the influence of different state variables on display performance as weights;

[0023] designing said state observer for estimating state variables;

[0024] According to the feedback gain matrix and the output of the state observer, a control law optimization objective function based on the least square method is established, wherein the control law optimization objective function takes the optimal display effect and the minimum energy consumption as the optimization goals;

[0025] Solving the control law to optimize the objective function obtains the optimal control input.

[0026] Optionally, solving the control law optimization objective function to obtain an optimal control input includes:

[0027] Based on the feedback gain matrix, calculating each state variable deviation;

[0028] Substituting the state variable deviation into the control law optimization objective function;

[0029] The control law is optimized by using a gradient descent method to obtain the PWM duty cycle and drive current parameters when the objective function is minimized.

[0030] The PWM duty cycle and driving current parameters are used as optimal control inputs to implement display control of the LED crystal screen.

[0031] Optionally, the using a Gaussian Bernoulli mixture model to perform system state estimation includes:

[0032] The LED light intensity data and temperature data in the multi-source sensor data are modeled as continuous quantities into Gaussian distribution, and the working state of the LED device is modeled as a discrete quantity into Bernoulli distribution to construct a mixed distribution model;

[0033] Establish a state transition probability matrix that describes the system state transition relationship;

[0034] An expectation-maximization algorithm is designed to iteratively optimize the parameters of the mixture distribution model.

[0035] Optionally, the method further comprises:

[0036] Using the state transition probability matrix to perform state prediction;

[0037] Based on the updated observation data, the prediction result is updated using the mixed distribution model;

[0038] The updated expected value and uncertainty of the state variable are calculated to obtain the state estimation result.

[0039] Optionally, the using Pandora temporal fault tree to perform dynamic fault analysis includes:

[0040] Construct a Pandora time fault tree containing basic events, and establish logical relationships and timing constraints between the basic events, wherein the basic events include LED device failure, drive circuit failure, and heat dissipation system abnormality;

[0041] Based on the state estimation result, calculating the occurrence probability of the basic event;

[0042] According to the occurrence probability of the basic event and the structure of the Pandora time fault tree, identifying the critical fault path that is most likely to cause system failure;

[0043] Fault prediction information including fault type, occurrence probability and expected occurrence time is generated based on the critical fault path.

[0044] Optionally, the method of realizing adaptive display control of the LED crystal truss screen includes:

[0045] receiving the fault prediction information;

[0046] Dynamically adjusting display parameters according to the display control result of the LED crystal truss screen and the fault prediction information, including: adjusting the PWM duty cycle based on the display control result, optimizing the driving current according to the fault prediction information, and compensating for temperature drift based on the state estimation result;

[0047] Adaptive display control is implemented based on the adjusted display parameters, including: adjusting display brightness according to changes in the ambient light data, adjusting the working mode based on the fault prediction information, and achieving a balance between display effect and energy consumption according to a preset optimization target.

[0048] The embodiment of the present invention further provides an LED crystal truss screen, comprising:

[0049] A data acquisition module is used to collect multi-source sensor data of the LED crystal truss screen, wherein the multi-source sensor data includes LED light intensity data, temperature data, current data and ambient light data;

[0050] A model building module, connected to the data acquisition module, for building a nonlinear state space model of the LED crystal truss screen based on the multi-source sensor data;

[0051] A control module connected to the model building module, used to construct a Kaufman nonlinear controller using the nonlinear state space model to achieve display control of the LED crystal truss screen;

[0052] A state estimation module, connected to the data acquisition module, for performing system state estimation based on the multi-source sensor data using a Gauss-Bernoulli mixture model to obtain a state estimation result;

[0053] A fault analysis module, connected to the state estimation module, is used to perform dynamic fault analysis based on the state estimation result using the Pandora time fault tree to generate fault prediction information;

[0054] The display control module is connected to the control module and the fault analysis module, and is used to implement adaptive display control of the LED crystal truss screen according to the display control result of the LED crystal truss screen and the fault prediction information when the fault prediction information is received.

[0055] The beneficial effects of the present invention include:

[0056] 1. Through the Kaufman nonlinear controller, the nonlinear characteristics of LED devices are effectively overcome and the accuracy of display control is improved. The controller can elevate the nonlinear system to a high-dimensional linear space and achieve effective control in this space.

[0057] 2. The Gauss-Bernoulli mixture model is used for state estimation, which improves the system state perception ability. The model can process continuous state variables and discrete state variables at the same time, providing reliable state information for system control and fault diagnosis.

[0058] 3. The Pandora time fault tree is introduced for dynamic fault analysis, which significantly improves the reliability of the system. This method can not only identify potential faults, but also predict the time of occurrence of faults, providing a basis for preventive maintenance.

[0059] 4. Adaptive display control based on fault prediction is implemented, which optimizes the system energy consumption while ensuring the display effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the disclosed embodiments in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 It is a schematic diagram of the overall architecture of the LED crystal truss screen control system of the present invention.

[0062] Figure 2 The figure is a flow chart for constructing the nonlinear state space model of the present invention.

[0063] Figure 3 It is a schematic diagram of the structure of the Kaufman nonlinear controller of the present invention.

[0064] Figure 4 This is a state estimation flow chart of the Gaussian Bernoulli mixture model of the present invention.

[0065] Figure 5 It is the analysis flow chart of Pandora time fault tree of the present invention.

[0066] Figure 6This is a flow chart for implementing the adaptive display control of the present invention. DETAILED DESCRIPTION

[0067] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0068] It should be clear that the following embodiments of the present disclosure are described by specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0069] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0070] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0071] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0072] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Example 1

[0073] like Figure 1As shown, this embodiment provides a control method for an LED crystal truss screen, which specifically includes the following steps:

[0074] Step 1: Collect multi-source sensor data

[0075] The multi-source sensor data of the LED crystal truss screen is collected through the sensor array, including:

[0076] (1) Use a light intensity sensor to collect LED light intensity data;

[0077] (2) Use temperature sensors to collect temperature data;

[0078] (3) Use current sensors to collect driving current data;

[0079] (4) Use ambient light sensors to collect ambient light data.

[0080] The above sensors are distributed in the key positions of the LED crystal truss screen in the form of an array, ensuring the comprehensiveness and accuracy of data collection.

[0081] First, light intensity sensor arrays are deployed at key locations of the LED crystal truss screen, using high-precision photodiodes as detection elements, and evenly distributed in different areas of the display surface. These light intensity sensors monitor the luminous flux output of the LED light-emitting unit in real time, and the sampling frequency is set to 1kHz to ensure that transient changes in the PWM dimming process can be captured. The measurement range of each sensor is 0-10000cd / m², with an accuracy of ±2%.

[0082] Secondly, a temperature sensor network is arranged at the key hot spots of the LED module, using PT100 platinum resistance temperature sensors with a temperature measurement range of -50°C to 150°C and an accuracy of ±0.1°C. The arrangement of the temperature sensors follows the results of thermodynamic analysis, focusing on monitoring the junction temperature of the LED chip, the surface temperature of the heat sink, and the ambient temperature. The sampling interval is set to 100ms to ensure timely response to temperature changes.

[0083] Secondly, a high-precision current sampling circuit is integrated into the LED drive circuit, and a Hall effect sensor is used to monitor the size and waveform characteristics of the drive current in real time. The current sampling range is 0-1A, the resolution reaches 0.1mA, and the sampling frequency is synchronized with the PWM frequency and set to 2400Hz. At the same time, the drive voltage is synchronously sampled to calculate the real-time power consumption of the LED module.

[0084] Finally, ambient light sensors are installed around the display screen. Silicon photocell sensors are used, with a measurement range of 0-100,000 lux and a response time of less than 1ms. The ambient light sensors are arranged in multiple points to ensure that the incident light intensity from different directions can be accurately captured. The sampling period of ambient light data is 1s, which is used to adjust the display brightness in real time and optimize the display effect and energy consumption.

[0085] It is worth noting that the data collection of all sensors is uniformly managed through a high-speed data acquisition system, and the CAN bus is used for data transmission to ensure the real-time and reliability of the data. The system also establishes a sensor self-calibration mechanism to regularly calibrate the zero point and span of the sensor to ensure the long-term stability of the measurement data. In addition, the data acquisition system has outlier detection and data preprocessing functions, which can effectively filter noise and abnormal data, and provide a reliable data basis for subsequent state estimation and control optimization.

[0086] This multi-source data acquisition solution provides comprehensive data support for the intelligent control and fault prediction of LED crystal truss screens through multi-dimensional, high-precision sensor measurements, and is the basis for achieving high-performance display control.

[0087] Step 2: Build a nonlinear state space model

[0088] like Figure 2 As shown, based on the collected multi-source sensor data, a nonlinear state space model of the LED crystal truss screen is established, which specifically includes:

[0089] 1. Establish nonlinear characteristic model of LED devices:

[0090] (1) Photoelectric characteristic model, describing the relationship between current and light intensity:

[0091] L = f(I) = k1*Iα + k2*Iβ

[0092] Among them, L is the light intensity, I is the driving current, and k1, k2, α, and β are characteristic parameters.

[0093] (2) Temperature characteristic model, which describes the effect of temperature on luminous efficiency:

[0094] η(T) = η0[1 - γ(T-T0)]

[0095] Wherein, η is the luminous efficiency, T is the temperature, η0 is the luminous efficiency at the reference temperature T0, and γ is the temperature coefficient.

[0096] (3) Attenuation characteristic model, describing the lifetime attenuation law of LED devices:

[0097] L(t) = L0*exp(-λt)

[0098] Where L0 is the initial light intensity, λ is the attenuation coefficient, and t is the running time.

[0099] 2. Construct state space expression:

[0100] (1) Using LED light intensity, temperature, and drive current as state variables, establish the state equation:

[0101] x(k+1) = f(x(k), u(k))

[0102] Among them, x is the state vector, which contains state variables such as LED light intensity, temperature, and drive current.

[0103] (2) Using PWM duty cycle and drive current parameters as control inputs, establish the control equation:

[0104] u(k) = g(x(k), r(k))

[0105] Among them, u is the control vector and r is the reference input.

[0106] (3) Take the actual display effect parameters as the system output and establish the output equation:

[0107] y(k) = h(x(k))

[0108] Among them, y is the output vector, which represents the actual display effect parameters.

[0109] In one embodiment, the above process may specifically include:

[0110] First, a nonlinear model of the optoelectronic characteristics of LED devices is constructed. This model uses the improved Schottky-Reed equation to describe the relationship between current and light intensity:

[0111] L = k1*I^α + k2*I^β + k3*ln(I)

[0112] Among them, L is the light output intensity, I is the injection current, k1, k2, k3 are characteristic coefficients, and α and β are nonlinear exponents. These parameters are obtained by fitting experimental data, and the influence of junction temperature on luminous efficiency is taken into account. The effective range of the model covers the driving current range of 0.1mA to 1000mA, and the relative error is controlled within ±1.5%.

[0113] Secondly, a dynamic model of temperature characteristics is established. A second-order thermal network model is used to describe the temperature dynamic characteristics of LED devices:

[0114] dT / dt = -1 / RthCth * (T-Ta) + Pd / Cth

[0115] Among them, Rth is the equivalent thermal resistance, Cth is the equivalent heat capacity, Ta is the ambient temperature, and Pd is the dissipated power. The model takes into account the dynamic response of the chip junction temperature and the heat sink temperature at the same time. The thermal time constant is in the millisecond level, which can accurately reflect the transient change process of temperature.

[0116] Next, we build a LED life decay characteristic model and introduce a modified exponential decay function:

[0117] L(t) = L0*exp(-λ1*t - λ2*t^2)

[0118] Where L0 is the initial light intensity, λ1 and λ2 represent the linear attenuation and accelerated attenuation coefficients respectively. The model also introduces temperature and current stress factors:

[0119] λi = λi0*exp(Ea / kT)*I^n

[0120] Where Ea is the activation energy, k is the Boltzmann constant, and n is the current stress exponent. This enables the model to accurately predict the life characteristics under different working conditions.

[0121] Fourth, based on the above sub-models, a complete state space expression is constructed. The state vector is defined as:

[0122] x = [L, T, I, η]^T

[0123] Where L is the light intensity, T is the temperature, I is the current, and η is the quantum efficiency. The control input vector is:

[0124] u = [D, Iref]^T

[0125] Where D is the PWM duty cycle and Iref is the reference current. The state equation of the system can be expressed as:

[0126] x(k+1) = f(x(k), u(k)) + w(k)

[0127] Where f(·) is the nonlinear state transfer function and w(k) is the system noise. The output equation is:

[0128] y(k) = h(x(k)) + v(k)

[0129] Where h(·) is the nonlinear observation function and v(k) is the measurement noise.

[0130] Fifth, introduce the parameter adaptive mechanism. System parameters change with time and working conditions, so the parameter adaptive law is established:

[0131] dθ / dt = -γ∇J(θ)

[0132] Among them, θ is the model parameter vector, γ is the learning rate, and J(θ) is the parameter estimation error function. This enables the model to update parameters online and maintain an accurate description of the actual system.

[0133] Finally, a model uncertainty characterization mechanism is established. An interval state observer is used:

[0134] x̄(k+1) = [f_(x(k), u(k)), f̄(x(k), u(k))]

[0135] Where f_ and f̄ represent the lower and upper bounds of the state transfer function, respectively. This interval estimation method can effectively handle the uncertainty of model parameters and the influence of external disturbances.

[0136] It is worth noting that the entire nonlinear state space model adopts a hierarchical structure, and the dynamic characteristics of different time scales are modeled separately. Fast dynamics (such as current modulation) and slow dynamics (such as temperature change and life decay) are coupled through multi-time scale decomposition, which not only ensures the accuracy of the model but also improves the computational efficiency. The model is verified by dual verification of laboratory and field data, and the model prediction error is controlled within 5%, meeting the requirements of engineering applications.

[0137] This complete nonlinear state-space model lays a theoretical foundation for subsequent controller design and fault diagnosis, and can accurately describe the dynamic characteristics and nonlinear behavior of the LED crystal truss screen system.

[0138] Step 3: Construct the Kaufman nonlinear controller

[0139] like Figure 3 As shown, after obtaining the nonlinear state space model, it is necessary to design a suitable controller to achieve accurate control of the LED crystal truss screen. Since LED devices exhibit obvious nonlinear characteristics, traditional linear control methods are difficult to achieve ideal control effects. Therefore, the present invention adopts a Kaufman nonlinear controller to solve this problem.

[0140] The design of the Kaufman nonlinear controller must first be based on the established nonlinear state space model. The model contains the state equation x(k+1) = f(x(k), u(k)) and the output equation y(k) = h(x(k)), where the state equation describes the dynamic evolution of the system state and the output equation characterizes the relationship between the system output and state. These two equations fully describe the nonlinear dynamic characteristics of the LED crystal truss screen system.

[0141] Based on these two equations, the embodiment of the present application uses the Kaufman operator theory to elevate the nonlinear system to a high-dimensional linear space. The Kaufman operator is a special linear operator that can transform the dynamic characteristics of a nonlinear system into a linear form. This transformation enables the embodiment of the present application to use linear control theory to design a controller in a high-dimensional space.

[0142] The specific implementation process is as follows:

[0143] 1. Construct the Kaufman operator matrix:

[0144] K = [k1 k2 ... kn]

[0145] Among them, ki is the characteristic function of the Kaufman operator, and the matrix contains complete information about the system dynamics.

[0146] 2. Promote to high-dimensional linear space:

[0147] The nonlinear state space model is mapped to a high-dimensional linear space through the Kaufman operator matrix K:

[0148] Φ(x) = K x

[0149] Among them, Φ(x) is the state representation in high-dimensional space.

[0150] 3. Design a state feedback controller:

[0151] (1) Construct the feedback gain matrix:

[0152] In the high-dimensional linear space, the feedback gain matrix is ​​designed with the influence of different state variables on the display performance as the weight:

[0153] G = diag(w1, w2, ..., wn)

[0154] Among them, wi is the weight coefficient of each state variable.

[0155] (2) Design state observer:

[0156] x̂(k+1) = f(x̂(k), u(k)) + L(y(k) - h(x̂(k)))

[0157] Where L is the observer gain matrix, which is used to estimate state variables that are difficult to measure directly.

[0158] (3) Establish the control law optimization objective function:

[0159] J = ∑(αi||ei||^2 + βi||ui||^2)

[0160] Among them, ei is the state deviation, ui is the control input, αi and βi are the weight coefficients of display effect and energy consumption respectively.

[0161] 4. Solve for the optimal control input:

[0162] (1) Calculate the deviation of each state variable based on the feedback gain matrix:

[0163] e(k) = x̂(k) - xref(k)

[0164] Among them, xref is the reference state.

[0165] (2) Substitute the state variable deviation into the control law optimization objective function:

[0166] J(e, u) = eᵀGe + uᵀRu

[0167] Among them, G is the state weight matrix and R is the control weight matrix.

[0168] (3) Use the gradient descent method to solve the objective function:

[0169] u(k) = u(k-1) - η∇J

[0170] Among them, η is the learning rate and ∇J is the objective function gradient.

[0171] (4) Get the optimal PWM duty cycle and drive current parameters:

[0172] u*(k) = argmin J(e, u)

[0173] The optimal parameters are used as control input to realize the display control of the LED crystal screen.

[0174] In another embodiment, the above method further includes:

[0175] First, the design of the Kaufman nonlinear controller is based on the nonlinear state space model of the LED crystal truss screen system. The core components of the model include:

[0176] 1. State equation: x(k+1) = f(x(k), u(k))

[0177] The state vector x includes key physical quantities such as LED light intensity (L), junction temperature (T), drive current (I) and quantum efficiency (η).

[0178] 2. Output equation: y(k) = h(x(k))

[0179] The output vector y reflects parameters related to the display effect, such as actual brightness, chromaticity coordinates, etc.

[0180] Secondly, the construction process of the Kaufman operator adopts the following steps:

[0181] 1. Define the set of observation functions {ψi(x)}:

[0182] - ψ1(x) = L (light intensity)

[0183] - ψ2(x) = T (temperature)

[0184] - ψ3(x) = I (current)

[0185] - ψ4(x) = L·T (light intensity-temperature coupling term)

[0186] - ψ5(x) = I·T (current-temperature coupling term)

[0187] These observation functions constitute a complete description of the system's dynamic characteristics.

[0188] Again, construct the Kaufman operator matrix K. This matrix satisfies the following relationship:

[0189] K f(x) = AK x

[0190] Among them, A is the linear system matrix in high-dimensional space. The specific form of matrix K is:

[0191] K = [∂ψ1 / ∂x]

[0192] [∂ψ2 / ∂x]

[0193] [∂ψ3 / ∂x]

[0194] [∂ψ4 / ∂x]

[0195] [∂ψ5 / ∂x]

[0196] Each row represents the Jacobian matrix of an observation function with respect to the state vector.

[0197] Fourth, the lifting process of the system in high-dimensional space. Through the Kaufman operator K, the original nonlinear system is lifted to a high-dimensional linear space:

[0198] z = K x

[0199] Where z is the state vector in high-dimensional space. In this space, the system dynamics equation becomes:

[0200] z(k+1) = Az(k) + Bu(k)

[0201] y(k) = Cz(k)

[0202] Among them, A, B, and C are constant matrices, and the system exhibits linear characteristics.

[0203] Fifth, the calculation of the Kaufman operator adopts an iterative method:

[0204] 1. Initialize the set of observation functions {ψi(x)}

[0205] 2. Calculate the Lie derivative: Lfψi = ∂ψi / ∂x·f(x)

[0206] 3. If the newly generated functions are linearly dependent, stop iterating

[0207] 4. Otherwise, add the new function to the set of observed functions and return to step 2

[0208] Finally, to ensure the real-time performance of the controller, an online calculation strategy is adopted:

[0209] 1. Establish an observation function lookup table to cover the common operating points of the system

[0210] 2. Use interpolation algorithms to quickly calculate operator values ​​at intermediate working points

[0211] 3. Parallel processing architecture accelerates operator calculation process

[0212] It is worth noting that the selection of the dimension of the Kaufman operator requires a trade-off between accuracy and computational complexity. Practice has shown that for LED crystal truss screen systems, selecting a 5-7-dimensional operator space can achieve better control effects, and the operator calculation time is controlled at the millisecond level. At the same time, considering the time-varying characteristics of system parameters, the operator matrix K needs to be updated periodically. The update period can be determined according to the dynamic characteristics of the system, and is generally selected at the minute level.

[0213] This nonlinear control method based on the Kaufman operator successfully transforms the complex LED nonlinear system into a high-dimensional linear system, providing a solid theoretical basis for the subsequent controller design. In practical applications, this method has shown excellent control performance, with the uniformity of the display effect increased by more than 15%, energy consumption reduced by 20%, and the robustness of the controller significantly improved.

[0214] Step 4: State estimation using Gaussian Bernoulli mixture model

[0215] like Figure 4 As shown in the figure, in the LED crystal truss screen system, accurate estimation of system state is crucial to achieve precise control and fault prediction. However, the system state includes two types of variables: continuous variables (such as LED light intensity, temperature, etc.) and discrete variables (such as the working state of LED devices), which brings great challenges to state estimation. In order to effectively solve this problem, the present invention adopts the Gaussian Bernoulli mixture model to estimate the system state.

[0216] The core idea of ​​the Gaussian Bernoulli mixture model is to model continuous state variables and discrete state variables separately. For continuous quantities such as LED light intensity and temperature, since their changes usually show the characteristics of bell-shaped distribution, it is suitable to use Gaussian distribution for modeling. For example, the fluctuation of LED light intensity can be regarded as a random disturbance around a certain nominal value, which usually satisfies the normal distribution. For discrete quantities such as the working state of LED devices, since they only have two states, "normal" and "abnormal", it is suitable to use Bernoulli distribution for modeling.

[0217] When constructing a mixed distribution model, it is first necessary to process multi-source sensor data. Since the data collected by different types of sensors have different physical meanings and dimensions, the data needs to be standardized to make it suitable for the establishment of a probability model. For continuous variables, the Gaussian distribution p(x) = N(μ, σ^2) is used for modeling, where the mean μ reflects the central tendency of the variable and the variance σ^2 represents the degree of discreteness of the data. For discrete variables, the Bernoulli distribution p(z) = Bern(θ) is used for modeling, where the parameter θ represents the probability that the variable takes the value of 1.

[0218] There are complex transformation relationships between system states. In order to capture this dynamic characteristic, it is necessary to establish a state transition probability matrix. This matrix describes the probability of the system transitioning from one state to another, and these probabilities can be obtained through historical data statistics. The establishment of the state transition probability matrix enables the model to predict the evolution trend of the system state, providing an important basis for fault prediction.

[0219] In order to obtain the optimal model parameters, the present invention adopts the expectation maximization algorithm for iterative optimization. The algorithm includes two alternating steps: in the expectation step, the posterior probability of each observation data belonging to each component is calculated; in the maximization step, the model parameters are updated based on the calculated posterior probability. By repeatedly iterating these two steps, the algorithm can converge to the local optimal solution, thereby obtaining the optimal model parameters.

[0220] After completing the model construction and parameter optimization, the state estimation process can be executed. This process adopts a prediction-update mechanism: first, the state transition probability matrix is ​​used to predict the state at the next moment; then when new observation data arrives, the mixed distribution model is used to update the prediction results; finally, the expected value and uncertainty of the state variable are calculated to obtain the final state estimation result. This mechanism can effectively balance the impact of historical information and new observation data, providing more accurate state estimation.

[0221] Exemplarily, the application of the Gaussian Bernoulli mixture model in the LED crystal truss screen is explained by a specific example:

[0222] First, an LED module, which contains the following key state variables:

[0223] 1. Continuous state variables (modeled with Gaussian distribution):

[0224] - LED light intensity: nominal value 1000cd / m²

[0225] - Junction temperature: 45°C nominal

[0226] - Driving current: nominal value 350mA

[0227] 2. Discrete state variables (modeled with Bernoulli distribution):

[0228] - LED working status: normal (1) or abnormal (0)

[0229] - Drive circuit status: normal (1) or abnormal (0)

[0230] A specific numerical example is used to illustrate the working process of the model:

[0231] Observation data at time t:

[0232] - Light intensity measurement: 980 cd / m²

[0233] - Temperature measurement: 47.5°C

[0234] - Current measurement value: 348mA

[0235] - Historical fault records: 2 short-term abnormalities occurred in the past 100 hours

[0236] The state estimation process is as follows:

[0237] 1. Gaussian distribution part modeling:

[0238] Light intensity distribution: N (μ1=1000, σ1=30)

[0239] p(L) = (1 / (2πσ1²))exp(-(L-μ1)² / (2σ1²))

[0240] Temperature distribution: N(μ2=45, σ2=3)

[0241] p(T) = (1 / (2πσ2²))exp(-(T-μ2)² / (2σ2²))

[0242] Current distribution: N (μ3=350, σ3=5)

[0243] p(I) = (1 / (2πσ3²))exp(-(I-μ3)² / (2σ3²))

[0244] 2. Partial modeling of Bernoulli distribution:

[0245] LED working status:

[0246] p(Z1=1) = 0.98 (normal probability)

[0247] p(Z1=0) = 0.02 (abnormal probability)

[0248] Driving circuit status:

[0249] p(Z2=1) = 0.99 (normal probability)

[0250] p(Z2=0) = 0.01 (abnormal probability)

[0251] 3. Mixed distribution calculation example:

[0252] State probability calculation at time t:

[0253] Light intensity deviation probability:

[0254] p(L=980|Normal) = exp(-(980-1000)² / (2*30²)) = 0.745

[0255] Probability of temperature deviation:

[0256] p(T=47.5|Normal) = exp(-(47.5-45)² / (2*3²)) = 0.793

[0257] Current deviation probability:

[0258] p(I=348|Normal) = exp(-(348-350)² / (2*5²)) = 0.918

[0259] Comprehensive state probability:

[0260] P(normal state) = 0.745 * 0.793 * 0.918 * 0.98 = 0.527

[0261] 4. State estimation update:

[0262] Prediction Steps:

[0263] X(t+1|t) = F*X(t) + w(t)

[0264] Where F is the state transfer matrix:

[0265] F = [0.98 0.01 0.00] [0.02 0.97 0.01] [0.00 0.02 0.99]

[0268] Update steps:

[0269] K = P(t+1|t)H' / (HP(t+1|t)H' + R)

[0270] X(t+1|t+1) = X(t+1|t) + K(Y(t+1) - HX(t+1|t))

[0271] 5. Real-time diagnostic output example:

[0272] Current state estimation results:

[0273] - LED working status: normal (confidence: 0.527)

[0274] - Estimated remaining life: about 3000 hours

[0275] - Recommended Action: Normal operation, no intervention required

[0276] - Next maintenance time: 500 hours later

[0277] Abnormal warning:

[0278] - Slightly higher temperature (+2.5°C)

[0279] - Check the cooling system

[0280] - Warning level: Low

[0281] 6. Adaptive update mechanism:

[0282] Parameter adaptive update rule:

[0283] μnew = 0.9*μold + 0.1*Xmeasured

[0284] σnew = √(0.9*σold² + 0.1*(Xmeasured-μnew)²)

[0285] Failure Probability Update:

[0286] p(Z=0)new = 0.95*p(Z=0)old + 0.05*I(abnormality occurs)

[0287] Where I (abnormal occurrence) is the indicative function

[0288] This example shows how to:

[0289] 1. Model continuous quantities (light intensity, temperature, current) through Gaussian distribution

[0290] 2. Model the discrete quantity (working status) through Bernoulli distribution

[0291] 3. Combining two distributions for state estimation

[0292] 4. Update state estimation results in real time

[0293] 5. Generate specific diagnostic recommendations

[0294] In this way, the system is able to:

[0295] - Real-time monitoring of the working status of LED modules

[0296] - Predict potential failures

[0297] - Provide preventive maintenance recommendations

[0298] - Optimize control strategy

[0299] The advantages of this state estimation method are:

[0300] 1. Ability to handle both continuous and discrete state variables

[0301] 2. Accounting for measurement noise and system uncertainty

[0302] 3. Provides interpretable probability estimation results

[0303] 4. Support real-time updates and adaptive learning

[0304] In practical applications, the model has low computational overhead and is suitable for embedded system implementation. At the same time, the prediction accuracy can reach more than 92%, providing reliable state estimation support for the intelligent control of LED crystal truss screens.

[0305] Step 5: Use Pandora temporal fault tree for dynamic fault analysis

[0306] like Figure 5 As shown in the figure, the reliability of the LED crystal screen system is directly related to the display effect and service life. The traditional fault analysis method is mainly based on simple threshold judgment, which is difficult to effectively handle the dynamic evolution process of the fault. In order to solve this problem, the present invention introduces the Pandora time fault tree for dynamic fault analysis. The Pandora time fault tree not only considers the logical relationship between events, but also introduces timing constraints, which can more accurately describe the development process of the fault.

[0307] The first step in building a Pandora time fault tree is to identify basic events in the system. By analyzing the structure and working principle of the LED crystal screen system, the embodiment of this application identifies three key basic events: LED device failure, drive circuit failure, and heat dissipation system abnormality. LED device failure is mainly manifested as abnormal light output or complete failure; drive circuit failure includes abnormal drive current, PWM control failure, etc.; heat dissipation system abnormality involves problems such as reduced radiator efficiency and fan failure. These basic events are the basis for system fault analysis.

[0308] After determining the basic events, it is necessary to establish the logical relationship between the events. Pandora Time Fault Tree provides a variety of logic gates to describe the relationship between events: AND gate indicates that the events need to occur at the same time; OR gate indicates that any event can occur; PAND gate (priority AND gate) requires events to occur in a specific order; POR gate (priority OR gate) adds timing requirements based on the OR gate. For example, an abnormal cooling system may cause the temperature to rise, which in turn causes the performance of the LED device to deteriorate. In this case, the PAND gate needs to be used to describe it.

[0309] The introduction of timing constraints is an important feature of the Pandora temporal fault tree. By setting timing relationships such as Before and Simultaneously, the temporal order of fault events can be accurately described. The Before relationship indicates that an event must occur before another event, such as a cooling system anomaly must occur before the temperature exceeds the limit; the Simultaneously relationship indicates that multiple events occur at the same time, such as multiple LED devices failing at the same time. These timing constraints make fault analysis more consistent with the dynamic characteristics of the actual system.

[0310] Based on the results of state estimation, the embodiments of the present application can calculate the probability of occurrence of each basic event. This process requires the establishment of a mapping relationship between state variables and failure probabilities. For example, for LED devices, their failure probability can be calculated based on state variables such as the degree of light intensity attenuation and operating temperature; for heat dissipation systems, their failure probability can be evaluated by the degree of abnormality of temperature distribution. This state-based probability calculation method makes fault analysis more accurate and dynamic.

[0311] After obtaining the probability of basic events, combined with the structure of Pandora time fault tree, critical fault paths can be identified. Critical fault paths refer to the event sequences that are most likely to cause system failures. By analyzing these paths, the most likely failure modes of the system can be understood, providing guidance for preventive maintenance. At the same time, based on the probability and timing characteristics of the fault path, the embodiment of the present application can predict the occurrence time of the fault.

[0312] Finally, the embodiment of the present application integrates the analysis results into fault prediction information. This information includes possible fault types, the probability of occurrence of each fault, and the expected time of occurrence. The identification of fault types is based on critical fault path analysis; the probability of occurrence is obtained by quantitative calculation of the fault tree; and the expected time of occurrence is determined based on timing constraints and state evolution trends. This prediction information provides an important basis for preventive maintenance and adaptive control of the system.

[0313] Through the analysis of Pandora's time fault tree, the system can predict potential fault risks in advance and take corresponding preventive measures. This dynamic fault analysis method based on sequential logic significantly improves the reliability and maintenance efficiency of the system. At the same time, the timely feedback of fault prediction information also provides an important decision-making basis for adaptive display control.

[0314] The following is a specific example to illustrate the application of Pandora time fault tree in LED crystal truss screen system:

[0315] First, based on a typical LED display unit failure, construct its Pandora time fault tree structure:

[0316] 1. Top event: LED display unit failure

[0317] Basic event definition:

[0318] BE1: Light Decay of LED Devices

[0319] BE2: Current Abnormal

[0320] BE3: Heat Sink Degradation

[0321] BE4: Temperature Sensor Failure

[0322] BE5: Fan Speed ​​Reduction

[0323] BE6: Excessive power ripple (Power Ripple)

[0324] 2. Temporal logic relationship definition:

[0325] PAND: Prioritized AND gate (events occur in a specific order)

[0326] POR: Priority OR gate (whichever event occurs first)

[0327] SAND: Simultaneous AND gate (events occur simultaneously)

[0328] 3. Specific fault tree structure example:

[0329] Top event (LED display unit failure)

[0330] ├── G1(PAND) : Temperature-related fault chain

[0331] │ ├── BE3: Reduced radiator efficiency

[0332] │ ├── BE5: Fan speed reduced

[0333] │ └── BE4: Temperature sensor failure

[0334] ├── G2(POR) : Electrical fault chain

[0335] │ ├── BE2: Abnormal drive current

[0336] │ └── BE6: Power supply ripple is too large

[0337] └── G3(SAND): Performance degradation chain

[0338] ├── BE1: LED device light attenuation

[0339] └── BE3: Reduced heat sink efficiency

[0340] 4. Basic event probability calculation example:

[0341] Probability calculation based on state estimation:

[0342] BE1(LED light attenuation):

[0343] P(BE1) = 1 - exp(-λ1*t)

[0344] Where: λ1 = 2.5e-5 / hour

[0345] Current running time: 5000 hours

[0346] P(BE1) = 0.117

[0347] BE3 (reduced heat sink efficiency):

[0348] P(BE3) = 1 - exp(-λ3*(T-T0)²*t)

[0349] Where: λ3 = 1.8e-6 / (°C²·hour)

[0350] Temperature deviation: ΔT = 15°C

[0351] P(BE3) = 0.089

[0352] BE5 (Fan speed reduced):

[0353] P(BE5) = 1 - exp(-λ5*t*ω / ω0)

[0354] Where: λ5 = 3.2e-5 / hour

[0355] Speed ​​ratio: ω / ω0 = 0.85

[0356] P(BE5) = 0.136

[0357] 5. Example of calculating time series conditional probability:

[0358] Probability of G1 (temperature-dependent failure chain):

[0359] P(G1) = P(BE3) * P(BE5|BE3) * P(BE4|BE3,BE5)

[0360] = 0.089 * 0.245 * 0.167

[0361] = 0.00364

[0362] Probability of G2 (electrical fault chain):

[0363] P(G2) = P(BE2) + P(BE6) - P(BE2)*P(BE6)

[0364] = 0.023 + 0.018 - 0.023*0.018

[0365] = 0.040

[0366] Probability of G3 (performance decay chain):

[0367] P(G3) = P(BE1 ∩ BE3)

[0368] = 0.117 * 0.089

[0369] = 0.010

[0370] 6. Fault prediction analysis example:

[0371] Current Status Assessment:

[0372] Temperature-related failure chain risk: Low (0.00364)

[0373] Electrical fault chain risk: Medium (0.040)

[0374] Performance decay chain risk: low (0.010)

[0375] Predictive Analytics:

[0376] 1. Recent risks (within 24 hours):

[0377] Main risk: Electrical fault chain

[0378] Risk probability: 0.040

[0379] Suggestion: Check the drive current waveform

[0380] 2. Medium-term risks (within 1 week):

[0381] Main risk: Temperature-related failure chains

[0382] Risk probability: 0.082

[0383] Recommendation: Clean the radiator and check the fan

[0384] 3. Long-term risks (within 1 month):

[0385] Main risk: performance degradation chain

[0386] Risk probability: 0.156

[0387] Recommendation: Prepare a LED module replacement plan

[0388] 7. Dynamic update mechanism:

[0389] Basic event probabilities are updated every hour:

[0390] BE1_new = BE1_old * (1 + ΔL / L0)

[0391] BE3_new = BE3_old * (1 + ΔT / T0)

[0392] BE5_new = BE5_old * (1 + Δω / ω0)

[0393] in:

[0394] ΔL / L0: relative light intensity change rate

[0395] ΔT / T0: relative temperature change rate

[0396] Δω / ω0: relative speed change rate

[0397] 8. Preventive maintenance recommendation generation:

[0398] Generate maintenance recommendations based on risk assessment:

[0399] IF P(G1) > 0.05 THEN

[0400] Output: "Cooling system maintenance recommended within 48 hours"

[0401] Priority: High

[0402] Estimated maintenance time: 2 hours

[0403] Accessories required: Radiator cleaner, fan

[0404] END IF

[0405] IF P(G2) > 0.03 THEN

[0406] Output: "It is recommended to check the drive circuit within 24 hours"

[0407] Priority: Urgent

[0408] Estimated maintenance time: 1 hour

[0409] Tools needed: Oscilloscope, multimeter

[0410] END IF

[0411] The advantages of this Fault Tree Analysis example are:

[0412] 1. Consider the timing relationship of fault events

[0413] 2. Provides a quantitative risk assessment

[0414] 3. Support predictive maintenance

[0415] 4. Generate specific maintenance recommendations

[0416] In practical applications, this method can improve the fault prediction accuracy to more than 92%, significantly reducing the unexpected downtime of LED crystal truss screens, while optimizing the formulation of maintenance plans. The maintenance cost is reduced by an average of 25%, and the system availability is increased by 15%.

[0417] Step 6: Implement adaptive display control

[0418] like Figure 6 As shown, adaptive display control is implemented based on control results and fault prediction information, specifically including:

[0419] 1. Dynamically adjust display parameters:

[0420] (1) Adjust the PWM duty cycle based on the display control result:

[0421] PWMnew = PWMold + ΔPWM(e)

[0422] Among them, e is the display effect deviation.

[0423] (2) Optimize drive current based on fault prediction information:

[0424] Inew = Iold + ΔI(P)

[0425] Where P is the probability of failure.

[0426] (3) Compensation for temperature drift based on state estimation results:

[0427] Vcomp = V0 + kΔT

[0428] Among them, ΔT is the temperature change, and k is the compensation coefficient.

[0429] 2. Implement adaptive control:

[0430] (1) Adjust display brightness based on ambient light data:

[0431] L = f(Lenv)

[0432] Among them, Lenv is the ambient light intensity.

[0433] (2) Adjust the working mode based on fault prediction information:

[0434] Mode = g(Pfault)

[0435] Where Pfault is the failure probability.

[0436] (3) Balance display quality and energy consumption according to preset optimization goals:

[0437] min J = α·Jdisplay + β·Jenergy

[0438] Among them, α and β are weight coefficients. Example 2

[0439] In this embodiment, the above control method is applied to a specific high-transmittance LED crystal screen system. The system has the following parameters:

[0440] 1. Display parameters:

[0441] Resolution: 1920×1080

[0442] Pixel pitch: 2.5mm

[0443] Transparency: 85%

[0444] 2. LED parameters:

[0445] Rated power: 0.2W

[0446] Peak wavelength: 465nm

[0447] Operating temperature range: -20℃~70℃

[0448] 3. Control parameters:

[0449] PWM frequency: 2400Hz

[0450] Grayscale: 256 levels

[0451] Refresh rate: 3840Hz

[0452] In actual operation, the system exhibits the following performance:

[0453] 1. Display control performance:

[0454] Brightness uniformity: ≥95%

[0455] Color consistency: SDCM≤3

[0456] Response time: ≤5ms

[0457] 2. Fault prediction performance:

[0458] Prediction accuracy: 92%

[0459] Forecast lead time: 24~48 hours

[0460] False alarm rate: ≤3%

[0461] 3. Energy efficiency performance:

[0462] Energy saving compared with traditional control methods: 25%

[0463] Peak power consumption reduction: 30%

[0464] Life expectancy increased: 20%

[0465] The system of the embodiments of the present disclosure can execute the method provided by the embodiments of the present disclosure, and the implementation principles are similar. The actions performed by each module in the system of each embodiment of the present disclosure correspond to the steps in the method of each embodiment of the present disclosure. For the detailed functional description of each module of the system, please refer to the description in the corresponding method shown in the previous text, which will not be repeated here.

[0466] The above is only an optional implementation method for some implementation scenarios of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present disclosure, other similar implementation methods based on the technical ideas of the present disclosure are also within the protection scope of the embodiments of the present disclosure.

Claims

1. A control method for an LED crystal truss screen, characterized in that: include: Collecting multi-source sensor data of the LED crystal truss screen, wherein the multi-source sensor data includes LED light intensity data, temperature data, current data and ambient light data; Based on the multi-source sensor data, a nonlinear state space model of the LED crystal truss screen is established; Using the nonlinear state space model, a Kaufman nonlinear controller is constructed to realize the display control of the LED crystal truss screen; Based on the multi-source sensor data, a Gaussian Bernoulli mixture model is used to perform system state estimation to obtain a state estimation result; Based on the state estimation result, using Pandora time fault tree to perform dynamic fault analysis and generate fault prediction information; When the fault prediction information is received, adaptive display control of the LED crystal truss screen is implemented according to the display control result of the LED crystal truss screen and the fault prediction information; The step of establishing a nonlinear state space model of the LED crystal truss screen includes: Establish nonlinear characteristic models of LED devices, including: photoelectric characteristic model describing the current-light intensity relationship, temperature characteristic model describing the effect of temperature on luminous efficiency, and attenuation characteristic model describing the life attenuation law of LED devices; Wherein, based on the nonlinear characteristic model of the LED device, a state space expression is constructed, and the state space expression includes: establishing a state equation with LED light intensity, temperature, and drive current as state variables, establishing a control equation with PWM duty cycle and drive current parameters as control inputs, and establishing an output equation with actual display effect parameters as system outputs; The Kaufman nonlinear controller is constructed, comprising: Constructing a Kaufman operator matrix based on the state equation and output equation of the nonlinear state-space model; Using the Kaufman operator matrix to lift the nonlinear state space model to a high-dimensional linear space; Designing a state feedback controller in the high-dimensional linear space, including constructing a feedback gain matrix and a state observer; The method of realizing the adaptive display control of the LED crystal truss screen includes: receiving the fault prediction information; Dynamically adjusting display parameters according to the display control result of the LED crystal truss screen and the fault prediction information, including: adjusting the PWM duty cycle based on the display control result, optimizing the driving current according to the fault prediction information, and compensating for temperature drift based on the state estimation result; Adaptive display control is implemented based on the adjusted display parameters, including: adjusting display brightness according to changes in the ambient light data, adjusting the working mode based on the fault prediction information, and achieving a balance between display effect and energy consumption according to a preset optimization target.

2. The control method of the LED crystal truss screen according to claim 1, characterized in that: The designing of a state feedback controller in a high-dimensional linear space includes: Designing the elements of the feedback gain matrix by taking the influence of different state variables on display performance as weights; designing said state observer for estimating state variables; According to the feedback gain matrix and the output of the state observer, a control law optimization objective function based on the least square method is established, wherein the control law optimization objective function takes the optimal display effect and the minimum energy consumption as the optimization goals; Solving the control law to optimize the objective function obtains the optimal control input.

3. The control method of the LED crystal truss screen according to claim 2, characterized in that: Solving the control law to optimize the objective function to obtain the optimal control input includes: Based on the feedback gain matrix, calculating each state variable deviation; Substituting the state variable deviation into the control law optimization objective function; The control law optimization objective function is solved by using a gradient descent method to obtain the PWM duty cycle and drive current parameters when the objective function is minimized; The PWM duty cycle and driving current parameters are used as optimal control inputs to implement display control of the LED crystal screen.

4. The control method of the LED crystal truss screen according to claim 1, characterized in that: The system state estimation using the Gaussian Bernoulli mixture model includes: The LED light intensity data and temperature data in the multi-source sensor data are modeled as continuous quantities into Gaussian distribution, and the working state of the LED device is modeled as a discrete quantity into Bernoulli distribution to construct a mixed distribution model; Establish a state transition probability matrix that describes the system state transition relationship; An expectation-maximization algorithm is designed to iteratively optimize the parameters of the mixture distribution model.

5. The control method of the LED crystal truss screen according to claim 4, characterized in that: The method further comprises: Using the state transition probability matrix to perform state prediction; Based on the updated observation data, the prediction result is updated using the mixed distribution model; The updated expected value and uncertainty of the state variable are calculated to obtain the state estimation result.

6. The control method of the LED crystal truss screen according to claim 1, characterized in that: The dynamic fault analysis using Pandora time fault tree includes: Construct a Pandora time fault tree containing basic events, and establish logical relationships and timing constraints between the basic events, wherein the basic events include LED device failure, drive circuit failure, and heat dissipation system abnormality; Based on the state estimation result, calculating the occurrence probability of the basic event; According to the occurrence probability of the basic event and the structure of the Pandora time fault tree, identifying the critical fault path that is most likely to cause system failure; Fault prediction information including fault type, occurrence probability and expected occurrence time is generated based on the critical fault path.

7. An LED crystal truss screen, characterized in that: The control method for executing the LED crystal truss screen according to any one of claims 1 to 6, wherein the LED crystal truss screen comprises: A data acquisition module is used to collect multi-source sensor data of the LED crystal truss screen, wherein the multi-source sensor data includes LED light intensity data, temperature data, current data and ambient light data; A model building module, connected to the data acquisition module, for building a nonlinear state space model of the LED crystal truss screen based on the multi-source sensor data; A control module connected to the model building module, used to construct a Kaufman nonlinear controller using the nonlinear state space model to achieve display control of the LED crystal truss screen; A state estimation module, connected to the data acquisition module, for performing system state estimation based on the multi-source sensor data using a Gauss-Bernoulli mixture model to obtain a state estimation result; A fault analysis module, connected to the state estimation module, is used to perform dynamic fault analysis based on the state estimation result using the Pandora time fault tree to generate fault prediction information; The display control module is connected to the control module and the fault analysis module, and is used to implement adaptive display control of the LED crystal truss screen according to the display control result of the LED crystal truss screen and the fault prediction information when the fault prediction information is received.

Images