Intelligent DLP seamless large screen dynamic adaptive adjustment system

Abstract

The invention provides an intelligent DLP seamless large-screen dynamic adaptive adjustment system, and relates to the technical field of DLP large-screen dynamic adaptive adjustment, and the system comprises a precision sensing unit, a heterogeneous calculation module, a dynamic compensation execution mechanism and a self-diagnosis subsystem. Optical, mechanical and environmental data are collected in real time through a backside illuminated CMOS sensor array, a dual-wavelength infrared photoelectric encoder and a distributed temperature sensor which are coaxially mounted; using an edge detection-optical flow fusion processor to identify the rainbow pattern, and combining with a time frequency feature extractor to quantify the stroboscopic intensity perceived by the human eyes; dynamic adaptive adjustment is realized through color wheel harmonic drive, LED nonlinear modulation and DMD time sequence compensation. The system integrates fault mode matching and a multi-objective optimization algorithm, the problems of rainbow lines, stroboscopic lag, chromatic aberration deviation and environmental interference of a traditional DLP system are effectively solved, and the display consistency and the user experience are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of DLP large-screen dynamic adaptive adjustment, and particularly to an intelligent DLP seamless large-screen dynamic adaptive adjustment system. Background Art

[0002] At present, with the continuous development of projection technology, DLP (Digital Light Processing) projection systems have been widely used in the field of seamless large-screen displays. However, there are still some problems in the actual use of existing DLP projection systems. For example, traditional DLP systems adopt a fixed color wheel rotation speed strategy, which is prone to red-green-blue color separation (rainbow pattern) in dynamic images, especially visible to the naked eye in low-refresh-rate scenarios. Existing flicker detection relies on single luminance sampling and cannot accurately quantify the flicker intensity perceived by the human eye, resulting in compensation lag. The color difference and timing deviation between multiple projection units lack systematic correction, affecting the consistency of large-screen displays. In the prior art, the color wheel system and color wheel control method proposed in Chinese Patent CN202310447128.1 do not consider the influence of ambient light interference on detection accuracy. The flicker suppression method in Chinese Patent CN202311718346.0 is only based on fixed threshold judgment and lacks modeling of human eye biometrics. Therefore, an intelligent DLP seamless large-screen dynamic adaptive adjustment system is proposed to solve the above problems. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide an intelligent DLP seamless large-screen dynamic adaptive adjustment system to solve at least the above problems.

[0004] The technical solution adopted by the present invention is as follows:

[0005] An intelligent DLP seamless large-screen dynamic adaptive adjustment system, the system includes:

[0006] Precision sensing unit: composed of a back-illuminated CMOS sensor array coaxially installed with the DLP optical engine, a dual-wavelength infrared photoelectric encoder, and a distributed temperature sensor, where the synchronization error between the exposure timing of the CMOS sensor and the flip clock of the DMD micromirror is <50ns;

[0007] Heterogeneous computing module: includes an edge detection-optical flow fusion processor for rainbow pattern recognition and a time-frequency feature extractor for flicker analysis, and the two share a data buffer through a PCIe Gen4×8 bus;

[0008] Dynamic compensation actuator: has a color wheel harmonic driver, an LED non-linear modulation circuit, and a DMD timing compensation engine;

[0009] Self-diagnosis subsystem: integrates a multi-spectral signal generator and a fault mode matching unit.

[0010] Furthermore, the edge detection - optical flow fusion processor executes a rainbow pattern detection method:

[0011] Apply the Canny operator with an adaptive threshold to each of the RGB channels, where the threshold for the red channel R_th = 0.7×I_red_max - 0.2×I_red_mean, I_red_max is the maximum gray value of the red channel, and I_red_mean is the average gray value;

[0012] Use the eight - direction gradient optical flow method to calculate the color band displacement vector. When the divergence angle θ of the red, green, and blue vector fields satisfies tanθ > 0.15 within a time window of ΔT = 20 ms, a rainbow pattern warning signal is generated;

[0013] Based on the theoretical relationship between the color wheel rotation speed ω and the color band spacing D: D = K·ω Λ (-1), where K is an optical system constant. Calibrate the value of K online by the recursive least - squares method. When the measured spacing deviation exceeds ±7% of the calibrated value three times continuously, compensation is triggered.

[0014] Furthermore, the time - frequency feature extractor includes:

[0015] A variable - length FFT unit that synchronously outputs the STFT time - frequency spectrum and the Wigner - Ville distribution;

[0016] The human eye perception stroboscopic index calculation model: HPFI = ∫[S(f)·W(f)·H(f)]df, where S(f) is the spectral energy density function, W(f) = e Λ (-0.002(f - 100) Λ 2) is the sensitivity window function, and H(f) = 1 / (1 + 0.5e Λ (-0.05f)) is the persistence effect attenuation factor;

[0017] A dynamic threshold adjustment mechanism: when the ambient illuminance > 500 lux, linearly increase the HPFI alarm threshold from 0.75 to 0.85.

[0018] Furthermore, the color wheel monitoring module in the precise perception unit includes:

[0019] The main detection channel: Decode the quadrature AB - coded signal and measure the rotation speed using the M / T method;

[0020] The auxiliary verification channel: Capture the laser speckle pattern through a 10 MSPS sampling ADC and calculate the angular velocity using the PIV algorithm;

[0021] Fault Arbiter: When the dual-channel difference > 3σ, start wavelet packet entropy analysis, where σ is the standard deviation of the angular velocity measurement values of the main detection channel and the auxiliary verification channel. When the entropy value > 4.5, a mechanical fault is determined; when the entropy value < 3.0, signal interference is determined.

[0022] Furthermore, the self-diagnosis subsystem performs:

[0023] Color wheel inertia test: Output test pulses with a linearly varying duty cycle from 10% to 100%. Report an error when the rise time t r > 2ms or the overshoot σ > 8%;

[0024] DMD micromirror diagnosis: Project a checkerboard pattern, and use the SUSAN algorithm with T = 12 and r = 3 pixels to statistically detect corner points. Mark the faulty area when the deviation > 3%;

[0025] System-level fault determination: Alarm when the normalized mutual information I(X; Y) / min(H(X), H(Y)) of the color wheel fault code and the DMD abnormal area > 0.7, where I(X; Y) is the mutual information of the color wheel fault code and the DMD abnormal area, and H(X), H(Y) are the information entropies of the color wheel fault code and the DMD abnormal area.

[0026] Furthermore, the dynamic compensation actuator includes:

[0027] Color wheel harmonic compensation: Inject a driving current I(t) = I0[1 + 0.05sin(6πf0t + φ)], where I0 is the reference driving current of the color wheel motor, f0 is the fundamental frequency of the color wheel, and the phase φ is adaptively adjusted according to the color band spacing;

[0028] LED pseudo-random blanking: Insert blanking pulses that follow a Weibull distribution;

[0029] DMD thermal compensation: Use the timing drift model Δt = 0.8ΔT + 0.02(ΔT) 2 to correct the clock edge in real time.

[0030] Furthermore, the fault mode matching unit includes:

[0031] First-level fusion: Calculate the fault belief BPA of temperature, vibration, and light intensity using the improved D-S evidence theory;

[0032] Second-level analysis: Construct a time Petri net model, and define the transition delay τ = 1 / (kΔT), where k is the thermal conductivity and ΔT is the temperature change;

[0033] Third-level prediction: Generate the future 5-second fault probability distribution through a bidirectional LSTM network. Start the degradation mode when P(fault) > 0.9.

[0034] Furthermore, it also includes a human factor optimization engine:

[0035] Bio - perception module: Monitor the change of pupil diameter with a 60Hz near - infrared camera, and synchronously collect the energy of EEG α - wave;

[0036] Comfort model: Q = 0.6(1 - ΔD / D0)+0.3EEG_α + 0.1HPFI, where ΔD is the change amount of the real - time pupil diameter, D0 is the reference pupil diameter, and EEG_α is the energy proportion of α - wave in the EEG signal;

[0037] Multi - objective optimizer: Use the NSGA - II algorithm to solve min(1 / Q, power consumption) and ΔE < 3, and output the Pareto - optimal parameter set.

[0038] Compared with the prior art, the beneficial effects of the present invention are:

[0039] 1. High - precision dynamic perception and compensation:

[0040] Adopt a back - illuminated CMOS sensor synchronized with the DMD micromirror (synchronization error < 50ns) and a dual - wavelength infrared encoder, combine the edge - detection - optical flow fusion algorithm (adaptive Canny threshold, eight - direction gradient optical flow method) to detect rainbow patterns in real time, and through color - wheel harmonic drive (phase adaptive adjustment) and DMD timing drift model (Δt = 0.8ΔT+0.02ΔT 2 ) for dynamic compensation to eliminate the influence of color separation and thermal drift.

[0041] 2. Optimization of human - eye perception of stroboscopic:

[0042] Based on the time - frequency feature extractor (variable - length FFT, Wigner - Ville distribution) and the human - eye perception stroboscopic index model (HPFI = ∫[S(f)·W(f)·H(f)]df), quantify the actual influence of stroboscopic on the human eye, dynamically adjust the alarm threshold (the threshold is increased to 0.85 when the ambient illuminance > 500lux), and combine with LED pseudo - random blanking (Weibull - distribution pulse) to significantly reduce the discomfort of the human eye.

[0043] 3. Multi - dimensional fault diagnosis and fault tolerance:

[0044] Through the main / auxiliary dual - channel color - wheel monitoring (M / T method, PIV algorithm) and wavelet packet entropy analysis (mechanical faults are determined when the entropy value > 4.5), combined with the self - diagnosis subsystem (color - wheel inertia test, DMD micromirror SUSAN algorithm detection) and normalized mutual - information determination (I(X;Y) / min(H(X),H(Y))>0.7), realize accurate fault location and system - level fault - tolerant switching.

[0045] 4. Human - factor optimization and comfort improvement:

[0046] Integrate a biological perception module (pupil diameter monitoring, EEG α-wave analysis) with a comfort model (Q = 0.6(1 - ΔD / D0) + 0.3EEG_α + 0.1HPFI), and use the NSGA-II algorithm to optimize the display parameters (Pareto optimal solution) to balance visual comfort and power consumption, ensuring the comfort of users during long-term viewing.

[0047] 5. Environment Adaptability and Real-time Performance:

[0048] Online calibrate the relationship between the color wheel rotation speed and spacing (K value) through the recursive least squares method, dynamically adjust the HPFI threshold in linkage with the ambient illuminance, and use the PCIe Gen4×8 bus to share data at high speed to ensure that the system response delay is less than 20ms, meeting the real-time adjustment requirements in complex environments. Brief Description of the Drawings

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only the preferred embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0050] Figure 1 It is a schematic diagram of the overall structure of an intelligent DLP seamless large screen dynamic adaptive adjustment system proposed in the embodiments of the present invention. Detailed Embodiments

[0051] In order to make the objectives, technical solutions, and advantages of the present invention more apparent, the following will describe in detail the exemplary embodiments according to the present invention with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited by the exemplary embodiments described here. Based on the embodiments of the present invention described herein, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0052] In the following description, a large number of specific details are given to provide a more thorough understanding of the present invention. However, it is obvious to those of ordinary skill in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some well-known technical features are not described.

[0053] It should be understood that the present invention can be implemented in different forms and should not be construed as limited to the embodiments presented here. On the contrary, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the present invention to those of ordinary skill in the art.

[0054] The purpose of the terms used herein is only to describe specific embodiments and is not a limitation of the present invention. As used herein, the singular forms "a", "an" and "the" are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the terms "comprising" and / or "including", when used in this specification, identify the presence of the stated features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups. As used herein, the term "and / or" includes any and all combinations of the associated listed items.

[0055] The principles and features of the present invention will be described below in conjunction with the accompanying drawings. The listed embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0056] Referring to Figure 1 , the present invention provides an intelligent DLP seamless large screen dynamic adaptive adjustment system, and the system includes:

[0057] Precision sensing unit: Composed of a back-illuminated CMOS sensor array coaxially installed with the DLP optical engine, a dual-wavelength infrared photoelectric encoder, and a distributed temperature sensor, wherein the exposure timing synchronization error of the CMOS sensor and the DMD micromirror flip clock is <50ns;

[0058] Heterogeneous computing module: Includes an edge detection-optical flow fusion processor for rainbow pattern recognition and a time-frequency feature extractor for stroboscopic analysis, and the two share a data buffer through a PCIe Gen4×8 bus;

[0059] Dynamic compensation actuator: Has a color wheel harmonic driver, an LED non-linear modulation circuit, and a DMD timing compensation engine;

[0060] Self-diagnosis subsystem: Integrates a multi-spectral signal generator and a fault mode matching unit.

[0061] Exemplarily, a precision sensing unit: coaxially installed with the DLP optical engine, collecting optical, mechanical, and environmental data in real time; a back-illuminated CMOS sensor array: coaxially installed at the rear end of the DLP optical engine, capturing the light intensity distribution of the RGB channels in the projection optical path at a sampling rate of 10 kHz. The sensor exposure timing is strictly synchronized with the DMD micromirror flipping clock, and the timing deviation controlled by the FPGA is less than 50 ns. A dual-wavelength infrared optoelectronic encoder: using 850 nm and 940 nm dual-wavelength laser sources, monitoring the color wheel rotation speed in real time through orthogonal AB encoded signals, and the encoded signals are transmitted to the main control unit after decoding. A distributed temperature sensor: PT1000 thermistors are arranged at the color wheel bearing, DMD chip, and LED light source, with a sampling frequency of 1 kHz and a temperature resolution of ±0.1 °C. A heterogeneous computing module: performing real-time analysis on the sensed data to identify anomalies such as rainbow patterns and stroboscopic; a dynamic compensation actuator: adjusting the operating parameters of the color wheel, LED, and DMD according to the calculation results; a self-diagnosis subsystem: realizing system health status monitoring through multi-spectral signal and fault mode matching; a human factor optimization engine: optimizing the display comfort by combining biological sensing data.

[0062] The edge detection - optical flow fusion processor executes the rainbow pattern detection method:

[0063] Apply the Canny operator with an adaptive threshold to each of the RGB channels, where the red channel threshold R_th = 0.7×I_red_max - 0.2×I_red_mean, I_red_max is the maximum gray value of the red channel, and I_red_mean is the average gray value;

[0064] Use the eight-direction gradient optical flow method to calculate the color band displacement vector. When the divergence angle θ of the red, green, and blue vector fields is detected to satisfy tanθ > 0.15 within a time window of ΔT = 20 ms, a rainbow pattern warning signal is generated;

[0065] Based on the theoretical relationship D = K·ω between the color wheel rotation speed ω and the color band spacing D Λ (-1), where K is an optical system constant, and the K value is calibrated online by the recursive least squares method. When the measured spacing deviation exceeds ±7% of the calibrated value three times in a row, compensation is triggered.

[0066] The time-frequency feature extractor includes:

[0067] A variable-length FFT unit, synchronously outputting the STFT time-frequency spectrum and the Wigner-Ville distribution;

[0068] The human eye perception stroboscopic index calculation model: HPFI = ∫[S(f)·W(f)·H(f)]df, where S(f) is the spectral energy density function, W(f) = e Λ (-0.002(f - 100) Λ2) is the sensitivity window function, and H(f) = 1 / (1 + 0.5e Λ (-0.05f)) is the decay factor of the persistence effect;

[0069] Dynamic threshold adjustment mechanism: When the ambient illuminance > 500 lux, the HPFI alarm threshold is linearly increased from 0.75 to 0.85.

[0070] Exemplarily, a variable-length FFT unit: synchronously outputs the short-time Fourier transform (STFT) spectrum and the Wigner-Ville distribution, with a resolution of 1 Hz.

[0071] Human eye perception flicker index (HPFI) model:

[0072] HPFI = ∫[S(f)·W(f)·H(f)]df, where S(f) is the spectral energy density function, and W(f) = e Λ (-0.002(f - 100) Λ 2) is the sensitivity window function, and H(f) = 1 / (1 + 0.5e Λ (-0.05f)) is the decay factor of the persistence effect.

[0073] Dynamic threshold adjustment: When the ambient illuminance > 500 lux, the HPFI alarm threshold is linearly increased from 0.75 to 0.85 to reduce the false alarm rate.

[0074] The color wheel monitoring module in the precision sensing unit includes:

[0075] Main detection channel: Decodes the quadrature AB encoded signal and measures the rotational speed using the M / T method;

[0076] Auxiliary verification channel: Captures the laser speckle pattern through a 10 MSPS sampling ADC and calculates the angular velocity using the PIV algorithm;

[0077] Fault arbiter: When the difference between the two channels > 3σ, starts wavelet packet entropy analysis, where σ is the standard deviation of the angular velocity measurement values of the main detection channel and the auxiliary verification channel. When the entropy value > 4.5, it determines a mechanical fault, and when the entropy value < 3.0, it determines signal interference.

[0078] Exemplarily, the main detection channel: Measures the rotational speed of the color wheel based on the M / T method, calculates the angular velocity by capturing the edge time difference of the AB encoded signal, with an accuracy of ±0.2%.

[0079] Auxiliary verification channel: Utilizes an ADC with a 10 MSPS sampling rate to capture the laser speckle pattern, combines the particle image velocimetry (PIV) algorithm to calculate the angular velocity of the color wheel, and compares it with the data of the main channel.

[0080] Fault Arbiter: When the dual-channel difference exceeds 3σ, start wavelet packet entropy analysis. If the entropy value > 4.5, it is determined as mechanical jamming; if the entropy value < 3.0, it is determined as signal interference, trigger an alarm and switch to the redundant channel.

[0081] The self-diagnosis subsystem performs:

[0082] Color wheel inertia test: Output test pulses with a linearly varying duty cycle from 10% - 100%. When the rise time t r > 2ms or the overshoot σ > 8%, report an error;

[0083] DMD micromirror diagnosis: Project a checkerboard pattern, use the SUSAN algorithm with T = 12 and r = 3 pixels to statistically count the corner points, and mark the fault area when the deviation > 3%;

[0084] System-level fault determination: When the normalized mutual information I(X; Y) / min(H(X), H(Y)) between the color wheel fault code and the DMD abnormal area > 0.7, trigger an alarm, where I(X; Y) is the mutual information between the color wheel fault code and the DMD abnormal area, and H(X), H(Y) are the information entropies of the color wheel fault code and the DMD abnormal area.

[0085] Exemplarily, output test pulses with a linearly varying duty cycle from 10% to 100%. If the rise time t r > 2ms or the overshoot σ > 8%, determine that the motor drive is abnormal.

[0086] Project a checkerboard pattern, use the SUSAN algorithm (threshold T = 12, radius r = 3 pixels) to statistically count the number of corner points. If the corner point deviation > 3%, mark the micromirror fault area.

[0087] Calculate the normalized mutual information between the color wheel fault code and the DMD abnormal area:

[0088] I(X; Y) / min(H(X), H(Y)) > 0.7. If the condition is met, trigger a system-level alarm and switch to the degraded mode.

[0089] The dynamic compensation actuator includes:

[0090] Color wheel harmonic compensation: Inject a drive current I(t) = I0[1 + 0.05sin(6πf0t + φ)], where I0 is the reference drive current of the color wheel motor, f0 is the fundamental frequency of the color wheel, and the phase φ is adaptively adjusted according to the color band spacing;

[0091] LED pseudo-random blanking: Insert blanking pulses that follow the Weibull distribution;

[0092] DMD thermal compensation: Use the timing drift model Δt = 0.8ΔT + 0.02(ΔT) 2 to correct the clock edge in real time.

[0093] Exemplarily, a harmonic drive current is injected into the color wheel motor: I(t) = I0[1 + 0.05sin(6πf0t + φ)], where the phase φ is adjusted according to the real-time color band spacing to compensate for the period error.

[0094] Blanking pulses that follow a Weibull distribution (shape parameter k = 2, scale parameter λ = 10 ms) are inserted to disperse the stroboscopic energy and reduce the human eye perception.

[0095] A timing drift model is established:

[0096] Δt = 0.8ΔT + 0.02(ΔT) 2 , where ΔT is the temperature rise of the DMD chip, and the micro-mirror flip clock edge is corrected in real time.

[0097] The fault mode matching unit includes:

[0098] First-level fusion: The fault belief degrees of temperature, vibration, and light intensity, i.e., BPA, are calculated using the improved D-S evidence theory.

[0099] Second-level analysis: A time Petri net model is constructed, and the transition delay τ = 1 / (kΔT) is defined, where k is the thermal conductivity and ΔT is the temperature change.

[0100] Third-level prediction: The future 5-second fault probability distribution is generated through a bidirectional LSTM network, and the degradation mode is activated when P(fault) > 0.9.

[0101] It also includes a human factor optimization engine:

[0102] Biological perception module: A 60 Hz near-infrared camera is used to monitor the change in pupil diameter, and the EEG α-wave energy is synchronously collected.

[0103] Comfort model: Q = 0.6(1 - ΔD / D0) + 0.3EEG_α + 0.1HPFI, where ΔD is the real-time change in pupil diameter, D0 is the reference pupil diameter, and EEG_α is the proportion of the energy of the α-wave in the EEG signal.

[0104] Multi-objective optimizer: The NSGA-II algorithm is used to solve min(1 / Q, power consumption) and ΔE < 3, and the Pareto optimal parameter set is output.

[0105] The above are only the preferred embodiments of the present invention, and they are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent DLP seamless large-screen dynamic adaptive adjustment system, characterized in that: The system comprises: Precision sensing unit: It consists of a back-illuminated CMOS sensor array coaxially mounted with the DLP optical engine, a dual-wavelength infrared photoelectric encoder, and a distributed temperature sensor. The CMOS sensor exposure timing and the DMD micromirror flip clock synchronization error are less than 50ns. Heterogeneous computing module: includes an edge detection-optical flow fusion processor for rainbow pattern recognition and a time-frequency feature extractor for stroboscopic analysis. The two share a data cache through a PCIe Gen4×8 bus. Dynamic compensation actuator: with color wheel harmonic driver, LED nonlinear modulation circuit and DMD timing compensation engine; Self-diagnosis subsystem: integrated multi-spectral signal generator and fault pattern matching unit.

2. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The edge detection-optical flow fusion processor performs a rainbow pattern detection method: The Canny operator with adaptive threshold is applied to the RGB channels respectively, where the red channel threshold R_th = 0.7×I_red_max-0.2×I_red_mean, I_red_max is the maximum gray value of the red channel, and I_red_mean is the average gray value; The eight-directional gradient optical flow method is used to calculate the color band displacement vector. When the divergence angle θ of the red, green and blue color vector field in the time window of ΔT = 20ms satisfies tanθ>0.15, a rainbow pattern warning signal is generated. Based on the theoretical relationship between the color wheel speed ω and the color band spacing D, D=K·ω Λ (-1), K is the optical system constant. The K value is calibrated online by the recursive least square method. The compensation is triggered when the measured spacing deviation exceeds ±7% of the calibration value for three consecutive times.

3. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The time-frequency feature extractor comprises: Variable length FFT unit, synchronous output of STFT time spectrum and Wigner-Ville distribution; The calculation model of human eye perception flicker index is: HPFI = ∫[S(f)·W(f)·H(f)]df, where S(f) is the spectrum energy density function, W(f) = e Λ (-0.002(f-100) Λ 2) is the sensitivity window function, H(f) = 1 / (1+0.5e Λ (-0.05f)) is the attenuation factor of the transient effect; Dynamic threshold adjustment mechanism: When the ambient illumination is >500 lux, the HPFI alarm threshold is linearly increased from 0.75 to 0.

85.

4. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The color wheel monitoring module in the precision sensing unit includes: Main detection channel: decodes orthogonal AB coded signals and uses M / T method to measure the rotation speed; Auxiliary verification channel: Capture the laser speckle pattern through a 10MSPS sampling ADC and apply the PIV algorithm to calculate the angular velocity; Fault arbitrator: When the dual-channel difference is greater than 3σ, wavelet packet entropy analysis is started, where σ is the standard deviation of the angular velocity measurement values ​​of the main detection channel and the auxiliary verification channel. An entropy value greater than 4.5 determines a mechanical fault, and an entropy value less than 3.0 determines a signal interference.

5. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The self-diagnostic subsystem performs: Color wheel inertia test: Output test pulses with a duty cycle linearly changing from 10% to 100%. r >2ms or overshoot σ>8% will cause an error; DMD micromirror diagnosis: Project a checkerboard pattern and use the SUSAN algorithm with T=12 and r=3 pixels to count corner points. If the deviation is >3%, mark the fault area. System-level fault judgment: When the normalized mutual information between the color wheel fault code and the DMD abnormal area I(X;Y) / min(H(X),H(Y))>0.7, an alarm is triggered, where I(X;Y) is the mutual information between the color wheel fault code and the DMD abnormal area, and H(X),H(Y) are the information entropy between the color wheel fault code and the DMD abnormal area.

6. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The dynamic compensation actuator comprises: Color wheel harmonic compensation: inject I(t)=I0[1+0.05sin(6πf0t+φ)] driving current, where I0 is the reference driving current of the color wheel motor, f0 is the base frequency of the color wheel, and the phase φ is adaptively adjusted according to the color band spacing; LED pseudo-random blanking: insert blanking pulses that follow Weibull distribution; DMD thermal compensation: using the timing drift model Δt = 0.8ΔT + 0.02(ΔT) 2 Correct the clock edge in real time.

7. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: The fault mode matching unit comprises: First-level fusion: use the improved DS evidence theory to calculate the fault confidence BPA of temperature, vibration, and light intensity; Second level analysis: construct a time Petri net model and define the transition delay τ = 1 / (kΔT), where k is the thermal conductivity and ΔT is the temperature change; Level 3 prediction: Generates the fault probability distribution in the next 5 seconds through a bidirectional LSTM network, and starts the degradation mode when P(fault)>0.

9.

8. The intelligent DLP seamless large-screen dynamic adaptive adjustment system according to claim 1, characterized in that: Also includes a human factors optimization engine: Biosensing module: uses a 60Hz near-infrared camera to monitor pupil diameter changes and simultaneously collect EEG α wave energy; Comfort model: Q = 0.6 (1-ΔD / D0) + 0.3EEG_α + 0.1HPFI, where ΔD is the real-time pupil diameter change, D0 is the reference pupil diameter, and EEG_α is the energy proportion of the α wave in the EEG signal; Multi-objective optimizer: Use NSGA-II algorithm to solve min(1 / Q, power consumption) and ΔE<3, and output the Pareto optimal parameter set.

Images