Artificial intelligence-based cinema projection equipment power management method and system

Abstract

The application belongs to the technical field of power management, and specifically relates to a cinema projection equipment power management method and system based on artificial intelligence, which comprises the following steps: collecting harmonic current and zero ground voltage disturbance signals of a cinema projection equipment power supply, obtaining a standard electric signal sequence through singular spectrum decomposition and phase space reconstruction; inputting the standard electric signal sequence into a double-domain phase synchronous mapping model, solving a manifold offset amount in combination with the characteristics of a chaotic system, and generating a disturbance characteristic vector; constructing a sparse Bayesian prediction structure according to the vector, nonlinearly estimating harmonic changes, and driving a chaotic oscillator to generate a leading compensation electric signal; injecting the signal into a power supply loop to realize disturbance cancellation, suppressing ground harmonic conduction to obtain an optimized power supply signal; collecting residual disturbance to calculate an exponential deviation and output a closed-loop correction instruction, updating mapping parameters and correcting a compensation track, and outputting a high-quality power supply signal. In the application, the disturbance is intelligently processed and closed-loop corrected, precise compensation is realized, stable power supply is ensured, and the reliable operation of the projection equipment is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of power management technology, specifically an artificial intelligence-based power management method and system for cinema projection equipment. Background Technology

[0002] The most critical drawback of traditional power management methods for cinema projection equipment is the lack of accurate disturbance identification and adaptive compensation capabilities. These methods often employ fixed-parameter filtering and static threshold suppression, failing to effectively analyze the complex dynamic characteristics of harmonic and ground voltage coupling disturbances. They also struggle to capture transient disturbances caused by projector load switching and power surges. Furthermore, the absence of a closed-loop correction mechanism prevents the quantification of compensation errors and real-time parameter adjustments, leading to issues such as compensation lag and low waveform matching. This results in incomplete harmonic suppression, ground loop interference propagation, insufficient power supply stability, and difficulty in maintaining the power supply signal within the equipment's permissible operating range over extended periods, ultimately impacting projection quality and equipment lifespan. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes an artificial intelligence-based power management method for cinema projection equipment. This invention primarily addresses the problems of power harmonic disturbances and power instability in projection equipment.

[0004] The present invention provides a power management method for cinema projection equipment based on artificial intelligence, comprising: S1: acquiring the harmonic current waveform and zero-ground voltage fluctuation signal of the power port of the cinema projection equipment, performing singular spectrum decomposition and phase space reconstruction on the original disturbance signal, and outputting a standard electrical signal sequence.

[0005] S2: Input the standard electrical signal sequence into the dual-domain phase synchronization mapping model, and use the synchronization deviation sensitivity of the chaotic system to solve the unstable manifold offset of harmonics and zero-ground disturbances in phase space, and generate the disturbance feature vector.

[0006] S3: Construct a sparse Bayesian phase prediction structure based on the perturbation feature vector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at future moments, drive the chaotic oscillator to generate a leading phase compensation waveform based on the prediction results, and output the compensation electrical signal.

[0007] S4: The compensation electrical signal is injected into the front-end power circuit of the projection equipment through the coupling structure. The disturbance and compensation signal are precisely canceled by nonlinear baseline fitting and alignment, the diffusion and propagation of harmonics in the grounding circuit are suppressed, and the optimized power supply signal is output.

[0008] S5: Based on the residual disturbance data of the optimized power supply signal, map it to the reconstructed phase space to calculate the Lyapunov exponential deviation, quantify the synchronization error between chaotic modulation and actual disturbance, and output closed-loop correction instructions.

[0009] S6: Update the coupling coefficient and manifold constraint of the dual-domain phase synchronization mapping according to the closed-loop correction instruction, correct the chaotic compensation trajectory in real time, so that the dynamic characteristics of the power supply signal remain stable within the working range allowed by the projection equipment, and output a high-quality power supply signal.

[0010] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for outputting the standard electrical signal sequence in step S1 are as follows: S11: Collect the harmonic current waveform and ground voltage fluctuation original disturbance signal of the power port of the cinema projection equipment, and perform equal-interval resampling and outlier removal, remove DC offset components, and construct one-dimensional original time series data.

[0011] S12: Construct a phase space trajectory matrix from the one-dimensional original time series data according to the preset embedding dimension and delay time. Through orthogonal decomposition of the trajectory matrix, separate the effective components and random noise components that characterize the disturbance features to obtain the core disturbance information.

[0012] S13: Sort and filter the singular values ​​of the core disturbance information, extract the main singular value components of the corresponding power disturbance, remove the random interference corresponding to the weak singular values, and obtain the reconstructed disturbance signal.

[0013] S14: Perform amplitude normalization processing on the reconstructed disturbance characteristic signal to eliminate dimensional differences and amplitude deviations, and arrange it into segments according to a fixed time window to form a standard electrical signal sequence.

[0014] According to the power management method for cinema projection equipment based on artificial intelligence provided by the present invention, the specific steps for generating the disturbance feature vector in step S2 are as follows: S21: Based on the power supply disturbance characteristics of cinema projection equipment, the standard electrical signal sequence is split into time-domain zero-ground voltage fluctuation and frequency-domain harmonic distribution, and the phase point trajectory and frequency domain characteristics that fit the load of the laser projector are constructed to form a dual-domain initial feature set.

[0015] S22: Substitute the initial features of the two domains into the chaotic dynamic equations that adapt to the perturbation bandwidth of the projection equipment for iterative mapping, and combine the instantaneous power impact characteristics to align the features with the chaotic attractors to obtain the phase space distribution results.

[0016] S23: Based on the synchronous deviation response of the chaotic system to the harmonics of the projection load and the zero-ground coupling disturbance, the displacement data of the trajectory deviating from the reference attractor is extracted from the phase space distribution, and the unstable manifold offset of the projection equipment disturbance is calculated.

[0017] S24: The unstable manifold offset and the dual-domain features are normalized according to the projection equipment perturbation weight and concatenated into a one-dimensional continuous vector with a fixed dimension, and the perturbation feature vector for power supply phase compensation in cinema projection is output.

[0018] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for obtaining the unstable manifold offset in step S23 are as follows: Locate the center of the ideal attractor in the phase space distribution results of the chaotic system, construct a local coordinate reference system, calculate the Euclidean distance between each phase point and the attractor center point by point, and form an initial displacement distribution sequence.

[0019] The initial displacement distribution sequence is nonlinearly smoothed to remove abrupt displacement values ​​and retain the trend displacement components coupled with harmonics and zero-ground disturbances, thus obtaining the displacement change sequence.

[0020] Based on the displacement change sequence, tangential and normal decompositions are performed along the phase space trajectory to solve for the displacement extrema and average offset rate in the two directions, thus obtaining the set of key offset parameters.

[0021] By substituting the set of key offset parameters into the manifold curvature calculation, the positional deviation and angular deflection of the disturbance on the unstable manifold are obtained, thus obtaining the unstable manifold offset corresponding to the power supply disturbance of the projection equipment.

[0022] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for outputting the compensation electrical signal in step S3 are as follows: S31: Based on the perturbation feature vector, the prior distribution of the sparse Bayesian model is initialized to construct a basis function space that adapts to the harmonic phase change. The feature dimension is reduced by automatic correlation determination to obtain the sparse Bayesian prediction model.

[0023] S32: Based on the sparse Bayesian prediction model and the historical dynamic laws in the perturbation eigenvector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at multiple future moments, and output the harmonic phase prediction sequence.

[0024] S33: Based on the harmonic phase prediction sequence and the amplitude fluctuation parameter in the disturbance characteristic vector, tune the iterative parameters and initial state of the chaotic oscillator to generate a chaotic compensation waveform sequence.

[0025] S34: Perform amplitude normalization and timing alignment calibration on the chaotic compensation waveform sequence to form a compensation electrical signal that perfectly matches the power supply disturbance of the cinema projection equipment.

[0026] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for outputting the optimized power supply signal in step S4 are as follows: S41: Collects the original compensation electrical signal of the power supply circuit at the front end of the cinema projection equipment, analyzes the harmonic components and noise distribution of the signal through an artificial intelligence time-series feature extraction algorithm, and outputs the feature-marked compensation electrical signal.

[0027] S42: The feature marker compensation electrical signal is injected into the front-end power circuit of the projection equipment through the coupling structure, the real-time disturbance waveform of the circuit is collected and the signal is superimposed, and the mixed electrical signal is output.

[0028] S43: The artificial intelligence nonlinear baseline fitting algorithm is used to perform baseline calibration and phase alignment on the mixed electrical signal, and the aligned signal is output.

[0029] S44: The alignment processing signal is introduced into the grounding loop conduction suppression stage, and the harmonic blocking model is used to suppress the propagation of harmonics in the grounding loop, and the harmonic-limited power supply signal is output.

[0030] S45: After optimizing the waveform and correcting the amplitude of the harmonic-limited power supply signal, verifying the cancellation effect and power supply stability, output the optimized power supply signal required by the cinema projection equipment.

[0031] According to the AI-based power management method for cinema projection equipment provided by the present invention, the specific steps for generating the closed-loop correction instruction in step S5 are as follows: S51: Collect and optimize residual disturbance data obtained from power supply signal retrieval, extract voltage amplitude and phase discrete values ​​point by point according to time sequence, remove abnormal sampling points and complete standardization processing, reconstruct phase points from one-dimensional time sequence array according to embedding dimension and delay time, map to high-dimensional space and generate phase space point set matrix.

[0032] S52: Calculate the local expansion rate of the trajectory based on the phase space point set matrix, traverse the entire time series to obtain the Lyapunov exponent numerical sequence, and perform point-by-point difference operation between the Lyapunov exponent numerical sequence and the preset chaotic modulation reference exponent to obtain the Lyapunov exponent deviation numerical sequence.

[0033] S53: Perform weighted statistics on the Lyapunov exponent deviation numerical sequence and calculate the mean square deviation value. Compare the deviation result with the preset threshold and output the synchronization error value.

[0034] S54: Based on the synchronization error value, convert it into amplitude and phase correction parameters through a linear mapping relationship, integrate the parameter information, and generate closed-loop correction instructions.

[0035] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for obtaining the Lyapunov exponent deviation numerical sequence in step S52 are as follows: Based on the phase space point set matrix, traverse each phase point in the matrix and select neighboring phase points. Calculate the Euclidean distance and time evolution difference between neighboring phase points and output the local expansion rate array corresponding to each phase point.

[0036] The local expansion rate array is subjected to time-weighted fitting, and the average exponential growth rate is calculated segment by segment along the trajectory evolution direction. The calculation results of the whole time period are integrated and normalized to output the Lyapunov exponent numerical sequence.

[0037] The Lyapunov exponent numerical sequence and the preset chaotic modulation benchmark exponent are calculated point by point according to the same time index. The difference result at each time point is recorded and the original difference sequence is output.

[0038] The original difference sequence is subjected to outlier removal and smoothing. After removing discrete jump points, it is rearranged according to time sequence to output the Lyapunov exponential deviation numerical sequence.

[0039] According to the artificial intelligence-based power management method for cinema projection equipment provided by the present invention, the specific steps for outputting a high-quality power supply signal in step S6 are as follows: S61: Perform discrete numerical analysis on the closed-loop correction instruction and complete the amplitude normalization operation, outputting the standardized instruction feature vector.

[0040] S62: Substitute the standardized instruction feature vector into the two-domain phase synchronization mapping iterative equation to complete the numerical update of the coupling coefficient and manifold constraint conditions, and output the mapping constraint parameter set.

[0041] S63: Perform point-by-point numerical correction on the chaotic compensation trajectory based on the mapping constraint parameter set, and output the numerical sequence of the compensation trajectory.

[0042] S64: Perform real-time numerical calibration and amplitude limiting filtering on the dynamic characteristics of the power supply signal based on the numerical sequence of the compensation trajectory, and output the numerical sequence of the power supply signal.

[0043] This invention also provides an artificial intelligence-based power management system for cinema projection equipment, comprising: The signal reconstruction module is used to acquire power supply harmonic current and zero-ground voltage fluctuation signals, process them through singular spectrum decomposition and phase space reconstruction, remove noise and normalize the timing, and output a standardized electrical signal sequence.

[0044] The phase domain solution module is used to input standard electrical signals into the dual-domain phase synchronization mapping model, solve the manifold offset based on the characteristics of the chaotic system, and generate a disturbance feature vector by fusing dual-domain features.

[0045] The phase prediction module is used to construct a sparse Bayesian model based on the perturbation feature vector, predict the harmonic amplitude and phase changes, and drive the chaotic oscillator to generate a leading phase compensation electrical signal.

[0046] The phase cancellation suppression module is used to inject the compensation signal into the power supply circuit, and achieve precise phase cancellation of disturbances through nonlinear baseline fitting, suppress the spread of harmonics in the grounding circuit, and output an optimized power supply signal.

[0047] The error quantization module is used to acquire residual disturbance data of the optimized signal, reconstruct the phase space and calculate the Lyapunov exponential deviation, quantify the synchronization error and generate closed-loop correction instructions.

[0048] The closed-loop correction module is used to update the mapping parameters and constraints according to the correction instructions, correct the chaotic compensation trajectory in real time, and output a high-quality power supply signal.

[0049] The artificial intelligence-based power management method for cinema projection equipment provided by this invention achieves precise compensation and stable power supply to ensure the reliable operation of the projection equipment by intelligently processing disturbances and performing closed-loop correction.

[0050] The beneficial effects of this invention are as follows: 1. This invention combines singular spectrum decomposition, phase space reconstruction, and chaotic dual-domain mapping with artificial intelligence time-series processing to collect, denoise, reconstruct, and extract features from harmonic current and zero-ground voltage fluctuation signals at the power supply port of cinema projection equipment. It transforms chaotic disturbances into quantifiable phase space trajectories and disturbance feature vectors, achieving a complete analysis of power supply disturbances from the time domain, frequency domain, and dynamic characteristics. This results in the accurate identification of minute disturbances, the elimination of environmental and equipment noise, and the improvement of disturbance feature recognition. Power supply disturbances no longer rely on manual threshold judgment and can adaptively capture transient changes caused by projector load switching and power surges, providing a highly reliable data foundation for subsequent compensation and avoiding compensation failure due to feature distortion.

[0051] 2. This invention links sparse Bayesian phase prediction with a chaotic oscillator, and completes nonlinear recursive estimation of harmonic amplitude and phase based on the disturbance feature vector. It dynamically tunes the chaotic compensation waveform and achieves timing and amplitude calibration, generating a compensation electrical signal that matches the disturbance in advance. This achieves the beneficial effects of predicting the disturbance trend in advance, realizing phase advance compensation, and improving waveform matching accuracy. Compared with the traditional fixed parameter compensation method, it can adapt to the changing operating conditions of cinema projection equipment, greatly reduce the phase shift problem caused by harmonics and zero-ground voltage fluctuations, and make the compensation signal highly aligned with the disturbance signal, laying the core foundation for accurate phase cancellation suppression. The sparsity of the model reduces the amount of computation and meets the real-time requirements of the projection site.

[0052] 3. This invention constructs a complete closed-loop control process from compensation, suppression, back-checking to command correction by using closed-loop disturbance retrieval, Lyapunov exponential deviation quantization, and dual-domain phase synchronous mapping for real-time correction. It continuously updates the coupling coefficient and manifold constraints and corrects the chaotic compensation trajectory, achieving the beneficial effects of dynamically eliminating residual disturbances, maintaining stable power supply signal dynamics, and locking power supply quality within the equipment's allowable operating range for a long time. It effectively suppresses harmonic propagation in the grounding loop, reduces the impact of power supply interference on projection quality and equipment lifespan, and achieves self-optimization and self-stabilization of power supply quality. It can adapt to long-term continuous projection scenarios without manual intervention, significantly improving the operational stability and intelligent power management level of cinema projection equipment. Attached Figure Description

[0053] The invention will now be further described with reference to the accompanying drawings.

[0054] Figure 1 This is a flowchart illustrating the steps of an artificial intelligence-based power management method for cinema projection equipment provided in an embodiment of the present invention. Figure 2 This is a flowchart of a power management method for cinema projection equipment based on artificial intelligence, provided in an embodiment of the present invention. Figure 3 This is a block diagram of an artificial intelligence-based power management system for cinema projection equipment provided in an embodiment of the present invention. Detailed Implementation

[0055] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below according to specific embodiments.

[0056] like Figures 1 to 3 As shown in the figure, the power management method for cinema projection equipment based on artificial intelligence provided in this embodiment of the invention includes: S1: Acquire the harmonic current waveform and zero-ground voltage fluctuation signal of the power port of the cinema projection equipment, perform singular spectrum decomposition and phase space reconstruction on the original disturbance signal, and output a standard electrical signal sequence.

[0057] S11: Collect the harmonic current waveform and ground voltage fluctuation original disturbance signal of the power port of the cinema projection equipment, and perform equal-interval resampling and outlier removal, remove DC offset components, and construct one-dimensional original time series data.

[0058] The harmonic current waveform and zero-ground voltage fluctuation signal at the power port of the cinema projection equipment were synchronously acquired. The sampling frequency was set to 10kHz-20kHz, and the sampling duration was no less than 10 seconds to ensure that the complete waveforms of the 5th and higher harmonics were captured, covering the typical period of the zero-ground voltage fluctuation. After acquisition, the original signal was resampled at equal intervals to uniformly adjust the time step of the original non-uniform sampling signal to 100μs. Linear interpolation was used to fill in the missing data points that appeared during the resampling process, ensuring the temporal continuity of the signal after resampling.

[0059] Outlier removal was performed using the Grubbs criterion with a significance level of 0.05. The mean and standard deviation of the original signal were calculated, and data points deviating from the mean by three times or more of the standard deviation were identified as outliers. Outliers were replaced by linear fitting of adjacent data points to avoid interference from outliers in subsequent processing. A moving window averaging method was used, with a window length of 100 sampling points. The moving average of the resampled signal was calculated to obtain the DC baseline value of the signal. The DC offset component was removed by subtracting the corresponding DC baseline value from the value of each sampling point in the original signal. The processed signal was then smoothed using a 5-point moving average filter. The average value of each sampling point and the two sampling points before and after it was taken and used to replace the value of the current sampling point, thus constructing a one-dimensional original time series data.

[0060] S12: Determine the optimal delay time using the mutual information method. Calculate the mutual information between the one-dimensional original time series data and the sequence after shifting by a certain step size. Set the shift step size to a range of 1-50 sampling points. Calculate the mutual information value corresponding to each shift step size. Select the shift step size corresponding to the first occurrence of the minimum mutual information value as the optimal delay time. The optimal delay time is set between 5-15 sampling points.

[0061] The optimal embedding dimension is determined using a false nearest neighbor algorithm, with a range of 2-10. For each candidate embedding dimension, the one-dimensional original time series data is mapped to a high-dimensional vector, and the proportion of false nearest neighbors for each high-dimensional vector is calculated. When the proportion of false nearest neighbors drops below 5%, the embedding dimension is determined as the optimal embedding dimension. Generally, the optimal embedding dimension is 3-6 dimensions. Based on the determined optimal embedding dimension *m* and optimal delay time *τ*, the one-dimensional original time series data is constructed into a phase space trajectory matrix. The matrix has N-(m-1)τ rows, where N is the total number of sampling points in the one-dimensional original time series data, and m columns. The element in the i-th row and j-th column of the matrix is ​​the value of the (i+(j-1)τ)-th sampling point in the one-dimensional original time series data. All elements are filled sequentially to complete the construction of the phase space trajectory matrix. Orthogonal decomposition is performed on the trajectory matrix using the singular value decomposition algorithm, decomposing the trajectory matrix into U, Σ, and V. T The product of three matrices, where U is the left singular vector matrix, Σ is the singular value diagonal matrix, and V...T The trajectory matrix is ​​a right singular vector matrix. Through orthogonal decomposition, the trajectory matrix is ​​split into multiple mutually orthogonal feature components. Each feature component corresponds to a singular value in the Σ matrix. Based on the magnitude of the singular value, the effective component with higher energy and the noise component with lower energy are initially separated. The feature component with a cumulative energy ratio of more than 95% is extracted as the core perturbation information.

[0062] S13: The eigenvalue decomposition method is used to calculate the singular values ​​corresponding to each eigencomponent, solving for the eigenvalues ​​of the covariance matrix of the core perturbation information. The calculated eigenvalues ​​are arranged in descending order, and each eigenvalue is the singular value of the corresponding eigencomponent. All singular values ​​are sorted in descending order of numerical value, and the eigencomponent corresponding to each singular value and its energy proportion in the original signal are recorded. The proportion of each singular value to the sum of all singular values ​​is calculated, i.e., the singular value contribution rate. An energy screening threshold of 85% is set, and the contribution rate of the sorted singular values ​​is accumulated. When the accumulated contribution rate reaches 85%, screening stops. All accumulated singular values ​​at this point are taken as principal singular values, and their corresponding eigencomponents are taken as principal singular value components. Singular values ​​with a remaining accumulated contribution rate below 15% are taken as weak singular values, and their corresponding eigencomponents are taken as weak singular value components. After screening, an inverse singular value reconstruction operation is performed, recombining the left singular vector, singular value, and right singular vector corresponding to the retained principal singular value components using U×Σ×V. T The inverse operation restores the high-dimensional feature components to the time domain signal. During the reconstruction process, linear interpolation is used to correct the edge distortion of the reconstructed signal to ensure that the beginning and end of the reconstructed signal are consistent with the trend of the original signal, thus obtaining the reconstructed disturbance signal. After reconstruction, the waveforms of the reconstructed signal and the core disturbance information are compared to ensure that the reconstructed signal can completely retain the core disturbance characteristics.

[0063] S14: The reconstructed disturbance signal is normalized using a maximum-minimum normalization algorithm. All sampling points of the reconstructed disturbance signal are traversed, and the maximum and minimum values ​​are recorded. Each sampling point in the reconstructed disturbance signal is mapped to a fixed numerical range of [0,1]. The normalization calculation for each sampling point is performed independently. After calculation, all normalized sampling points are traversed to check for outliers exceeding the [0,1] range. If any exist, a clamping method is used to correct the outliers to 0 or 1. Signal segmentation is then performed to analyze the fluctuation period of the reconstructed disturbance signal. The characteristic period T of the signal is calculated using the autocorrelation function. The time window width is set to 1.5T, and the sliding step size of the time window is set to 0.5T. Overlapping segmentation is used to segment the continuous normalized signal. For each segment, the starting and ending sampling points are recorded. After segmentation, the length of each signal segment is regularized, and the number of sampling points in each segment is uniformly adjusted to a fixed value. For segments with fewer than the fixed number of sampling points, zero-padding is used to fill in the gaps to the fixed length. For segments exceeding the fixed number of sampling points, the first 1024 sampling points are retained using a truncation method. All regularized signal segments are then arranged sequentially according to their original timing sequence to form a standard electrical signal sequence. Each segment in the sequence contains complete disturbance characteristics, and the amplitude and length are uniform.

[0064] S2: Input the standard electrical signal sequence into the dual-domain phase synchronization mapping model, and use the synchronization deviation sensitivity of the chaotic system to solve the unstable manifold offset of harmonics and zero-ground disturbances in phase space, and generate the disturbance feature vector.

[0065] S21: Based on the power supply disturbance characteristics of cinema projection equipment, the standard electrical signal sequence is split into two dimensions: the time-domain zero-ground voltage fluctuation pattern and the frequency-domain harmonic distribution. The time-domain phase point trajectory and the frequency-domain harmonic characteristic distribution of the projection power supply are respectively constructed to fit the load of the laser projector, forming a dual-domain initial feature set adapted to the cinema scene.

[0066] For standard electrical signal sequences that have undergone singular spectrum decomposition and phase space reconstruction, a two-dimensional partitioning is performed based on the dynamic characteristics of typical power supply disturbances in cinema projection equipment. The continuously changing signal components are divided into time-domain branches, and the components reflecting the spectral structure are divided into frequency-domain branches. In the time domain, the rising edge, falling edge, amplitude, and periodic fluctuations of the zero-ground voltage fluctuation are used as key features. Through a sliding phase point construction method, the one-dimensional time-series signal is expanded into a discrete set of phase points with a fixed delay time and embedding dimension, generating a time-domain phase point trajectory that reflects the voltage fluctuation trend, ensuring that the trajectory shape is consistent with the dynamic characteristics of the laser projector's power supply fluctuations. In the frequency domain, a piecewise Fourier transform is performed on the same sequence to extract the amplitude and phase information of typical low-frequency harmonic components (3rd, 5th, and 7th orders) and high-frequency spurious components. A frequency-domain feature distribution is constructed with the harmonic order as the abscissa and the normalized amplitude as the ordinate. Background noise components unrelated to the projection equipment's power supply are eliminated, while effective spectral lines characterizing harmonic distortion and ground coupling interference are retained. The generated time-domain phase point trajectories and frequency-domain harmonic characteristic distributions are combined, and their data lengths and sampling references are unified to form a dual-domain initial feature set that complements and constrains each other in the time and frequency domains. This set fully preserves the zero-ground voltage fluctuations and time-domain transient characteristics, as well as the harmonic distributions and frequency-domain distortion characteristics, and can comprehensively reflect the complete dynamic form of power supply disturbances in cinema projection equipment.

[0067] S22: Substitute the initial feature set of the two domains into the chaotic dynamic equation that adapts to the disturbance bandwidth of the cinema projection equipment for iterative mapping. Combined with the instantaneous power impact characteristics of the projector, the two domain features are gradually aligned with the trajectory of the chaotic attractor to obtain the phase space distribution result under the synchronous state that reflects the disturbance law of the power supply of the cinema projection.

[0068] The initial set of dual-domain features is used as the initial state and substituted into the chaotic dynamic equations pre-tuned according to the disturbance bandwidth range of the cinema projection equipment for iterative calculation. The control parameters of the chaotic system are calibrated based on the typical operating frequency band of the projector, the instantaneous power jump amplitude, and the common fluctuation range of zero-ground voltage. During the iteration process, the time-domain phase point trajectory is used as the state evolution constraint of the chaotic system, and the frequency-domain harmonic characteristic distribution is used as the amplitude boundary constraint of the system, so that the dual-domain features gradually approach the chaotic attractor trajectory in the iteration. After each iteration, the synchronization error between the current state and the ideal attractor trajectory is calculated, and the iteration step size is fine-tuned through error feedback until the synchronization error is reduced to within the preset threshold range. After the iteration is completed, all converged state points are arranged in evolution order to form a phase space distribution result that can completely reflect the nonlinear law of the power supply disturbance of the cinema projection.

[0069] S23: Based on the synchronous deviation response of the chaotic system to the harmonics and zero-ground coupling disturbance of the cinema projection load, the displacement data of the trajectory deviating from the reference attractor is extracted from the phase space distribution results, and the unstable manifold offset corresponding to the harmonics and zero-ground disturbance specific to the cinema projection equipment is calculated.

[0070] Taking the phase space distribution results of the synchronized state as the analysis object, the ideal attractor center of the chaotic system under undisturbed conditions is located in the phase space. A local coordinate reference system is constructed using this center as the origin. All actual evolution trajectory points in the phase space are traversed, and the Euclidean distance between each phase point and the attractor center is calculated point by point, forming an initial displacement distribution sequence characterizing the degree of trajectory deviation. This initial displacement distribution sequence is then nonlinearly smoothed to remove abrupt displacement jumps introduced by random noise, retaining only the trend displacement components caused by harmonic and zero-ground coupling disturbances, resulting in a normalized displacement change sequence. Based on this displacement change sequence, orthogonal decomposition is performed along the tangent and normal directions of the phase space trajectory, solving for the displacement extrema and average offset rate in both directions, combining them to form a set of key offset parameters containing amplitude, direction, and trend of change. By substituting the key offset parameter set into the manifold curvature calculation formula and combining the geometric definition of the unstable manifold, the curvature of the offset feature matrix is ​​calculated and the topological structure is analyzed. The expansion direction and angular deflection of the disturbance trajectory in the tangent space of the manifold are solved. Then, the overall trend of the manifold offset is determined by least squares fitting. The multi-dimensional offset information is fused and dimensionless to eliminate the difference in numerical magnitude, and a unified characterization of the unstable manifold offset of the power supply disturbance of the projection equipment is obtained.

[0071] S24: The unstable manifold offset and the dual-domain features are normalized and spliced ​​according to the perturbation weight of the cinema projection equipment, and arranged in a fixed dimension according to the analysis requirements of the projection power supply features to form a one-dimensional continuous vector, and the perturbation feature vector of the cinema projection power supply phase compensation is output.

[0072] The unstable manifold offset is fused with the initial feature set of the two domains. Feature weights are assigned according to the importance of the power supply disturbance to the cinema projection equipment's operation. Time-domain fluctuation features, frequency-domain harmonic features, and manifold offset are each assigned a corresponding weighting coefficient, highlighting the disturbance components closely related to projector stability. Extremum normalization is performed on the weighted feature data sets to map all feature values ​​to a unified interval, eliminating computational imbalances caused by differences in dimensions and numerical ranges. After normalization, various features are sequentially concatenated according to a preset fixed dimension. The time-domain feature sequence is arranged first, followed by the frequency-domain feature sequence, with the unstable manifold offset placed at the end of the sequence as a higher-order dynamic feature, forming a one-dimensional continuous vector with unified dimension, fixed order, and complete information. During the concatenation process, duplicate information and redundant components are removed, and the correlation between features is orthogonalized to reduce information coupling within the vector. The resulting disturbance feature vector fully encompasses the zero-ground voltage fluctuation pattern, harmonic distribution structure, and chaotic manifold offset characteristics, comprehensively and compactly describing the core dynamic characteristics of power supply disturbances in cinema projection equipment.

[0073] S3: Construct a sparse Bayesian phase prediction structure based on the perturbation feature vector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at future moments, drive the chaotic oscillator to generate a leading phase compensation waveform based on the prediction results, and output the compensation electrical signal.

[0074] S31: Based on the perturbation feature vector, the prior distribution of the sparse Bayesian model is initialized to construct a basis function space that adapts to the harmonic phase change. The feature dimension is reduced by automatic correlation determination to obtain the sparse Bayesian prediction model.

[0075] First, the acquired power supply disturbance feature vectors are extracted, and their internal feature parameters are used as the input basis for the sparse Bayesian model. Prior distribution initialization is performed according to the model parameter distribution characteristics. During initialization, a basis function space adapted to the dynamic characteristics of harmonic phase changes in the power supply of cinema projection equipment is constructed. The form and dimensions of the basis functions closely match typical change patterns such as harmonic phase shifts and phase jumps, avoiding subsequent fitting biases due to function space mismatch. The disturbance feature vectors are then imported into an automatic correlation determination module. This module quantifies the correlation between each feature dimension and harmonic phase changes, eliminating redundant dimensions with correlations below a set threshold, and retaining effective feature dimensions that significantly drive harmonic phase changes, thus achieving feature dimension simplification and compression. After dimension simplification, the model's weight parameters and noise variance parameters are iteratively updated by combining the optimized prior distribution and the filtered feature data, completing the overall construction of the sparse Bayesian prediction model.

[0076] S32: Based on the sparse Bayesian prediction model and the historical dynamic laws in the perturbation eigenvector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at multiple future moments, and output the harmonic phase prediction sequence.

[0077] Using a sparse Bayesian prediction model as the core inference vehicle, historical power supply dynamics data contained in the disturbance feature vector are imported. This data covers the dynamic patterns corresponding to the amplitude and phase fluctuation trajectory of continuous power supply disturbances and equipment load switching. Based on the probability distribution results output by the model, a nonlinear recursive estimation framework is constructed in conjunction with historical dynamic patterns. Taking the harmonic phase state at the current moment as the starting point for recursion, the amplitude and phase changes are predicted and calculated moment by moment according to the time series characteristics of the power supply disturbance of the cinema projection equipment. The posterior probability weights of the model output are introduced during the recursion process to correct the phase offset error caused by nonlinear fitting. For multi-moment prediction requirements, a time step matching the equipment power supply sampling frequency is set to sequentially calculate the harmonic amplitude and phase values ​​for the next moment and multiple future moments. Each recursive step uses the prediction result of the previous moment as the input condition for the current moment, maintaining the continuity of the temporal logic. After the recursive calculation is completed, the phase data predicted at each moment are integrated and arranged in chronological order to form a continuous harmonic phase prediction sequence.

[0078] S33: Based on the harmonic phase prediction sequence and the amplitude fluctuation parameter in the disturbance characteristic vector, tune the iterative parameters and initial state of the chaotic oscillator to generate a chaotic compensation waveform sequence.

[0079] The power supply amplitude fluctuation parameters, including amplitude fluctuation range, fluctuation frequency, and fluctuation mutation threshold, are extracted from the disturbance feature vector using the harmonic phase prediction sequence as the core. These parameters are then imported into the chaotic oscillator parameter tuning module. Based on the temporal variation pattern of the phase prediction sequence and the numerical range of the amplitude fluctuation parameters, the module tunes the core iterative parameters of the chaotic oscillator, such as the iteration step size, bifurcation parameter, and coupling coefficient. The initial state of the oscillator is set according to the initial phase value of the phase prediction sequence to ensure that the oscillator's operating state highly matches the phase and amplitude variation patterns of the power supply disturbance. After parameter tuning, the chaotic oscillator iterative calculation is initiated. The oscillator generates a chaotic time-series signal according to the tuned parameters and initial state. The phase change trajectory of the signal matches the harmonic phase prediction sequence, and the amplitude change trend matches the amplitude fluctuation characteristics of the power supply disturbance. The iteration process continues until time-series data covering the complete disturbance cycle is generated. This set of time-series data is then processed to form a chaotic compensation waveform sequence.

[0080] S34: Perform amplitude normalization and timing alignment calibration on the chaotic compensation waveform sequence to form a compensation electrical signal that perfectly matches the power supply disturbance of the cinema projection equipment.

[0081] Amplitude normalization is performed on the chaotic compensation waveform sequence. Based on the rated amplitude range of the power supply voltage of the cinema projection equipment, the amplitude extrema of the waveform sequence are calculated. A linear scaling algorithm is used to uniformly map the waveform amplitude to the standard amplitude range of the equipment power supply, eliminating amplitude overflow or underestimation caused by oscillator iteration, and ensuring that the amplitude of the compensation waveform is on the same order of magnitude as the amplitude of the actual power supply disturbance. Timing alignment calibration is performed by extracting the sampling timestamps of the power supply disturbance feature vector and the timing nodes of the phase prediction sequence. Using the initial moment of the disturbance as the reference point, the timing nodes of the chaotic compensation waveform sequence are matched point-by-point with the actual timing nodes of the power supply disturbance, correcting timing offsets and phase lags in the waveform sequence, ensuring that the compensation waveform is completely synchronized with the power supply disturbance in the time dimension. During the calibration process, fine-tuning is performed on the phase deviation of the waveform to ensure that the phase change nodes of the compensation waveform correspond precisely to the phase jump nodes of the power supply disturbance. After completing the dual calibration of amplitude normalization and timing alignment, the processed waveform sequence is converted into a continuous electrical signal, forming a compensation electrical signal that is perfectly matched with the power supply disturbance of the cinema projection equipment in terms of amplitude, phase, and timing.

[0082] S4: The compensation electrical signal is injected into the front-end power circuit of the projection equipment through the coupling structure. The disturbance and compensation signal are precisely canceled by nonlinear baseline fitting and alignment, the diffusion and propagation of harmonics in the grounding circuit are suppressed, and the optimized power supply signal is output.

[0083] S41: Acquires the raw compensation electrical signal from the front-end power circuit of the cinema projection equipment. It analyzes the harmonic components and noise distribution of the signal using an artificial intelligence time-series feature extraction algorithm, outputting a feature-labeled compensation electrical signal. The acquisition process employs a high-precision signal acquisition module controlled by artificial intelligence to capture the raw compensation electrical signal from the front-end power circuit of the projection equipment in real time. The acquisition range covers all operating conditions, including standby, normal projection, image switching, and high-brightness playback, while simultaneously recording related data such as signal acquisition time and equipment operating parameters. After acquisition, the raw time-domain electrical signal is converted into a high-dimensional time-series feature vector and input into the artificial intelligence time-series feature extraction network. This network, trained and optimized with a large amount of cinema power supply disturbance data, can adaptively identify harmonic components in different frequency bands, accurately distinguish between the fundamental wave and odd harmonics, and effectively separate redundant components such as random noise, power frequency interference, and equipment electromagnetic radiation. Through a fusion model of convolutional neural network and long short-term memory network, the amplitude, phase, occurrence time, and duration of each harmonic component are automatically labeled, quantifying the noise energy density. The feature labeling process strictly follows the original amplitude and phase relationship of the signal, and assigns each component a unique classification label and weight parameters only through artificial intelligence logic, clarifying the perturbation type and compensation priority corresponding to different features.

[0084] S42: The feature marker compensation electrical signal is injected into the front-end power circuit of the projection equipment through the coupling structure, the real-time disturbance waveform of the circuit is collected and the signal is superimposed, and the mixed electrical signal is output.

[0085] A non-inductive isolation coupling structure combining magnetic and capacitive coupling is used to avoid load impact on the original power supply circuit during injection. An AI timing synchronization module controls the injection timing, ensuring time-domain synchronization between the compensation signal and the circuit disturbance signal by acquiring the circuit clock signal and equipment operating status, preventing phase shift. After injection, a high-precision acquisition module is activated to acquire real-time voltage and current disturbance waveforms of the circuit according to the equipment's operating frequency band, capturing harmonics and fluctuations caused by the switching power supply, dimming module, etc., and recording the superposition process parameters. The acquired disturbance waveform and the injected characteristic-marked compensation electrical signal are superimposed point-to-point, completely preserving all signal details without any filtering or simplification. After superposition, an AI verification module checks the signal integrity, confirming that the compensation signal is undistorted, attenuated, and effectively integrated into the circuit, outputting a mixed electrical signal containing both disturbance and compensation components.

[0086] S43: The artificial intelligence nonlinear baseline fitting algorithm is used to perform baseline calibration and phase alignment on the mixed electrical signal, and the aligned signal is output.

[0087] The mixed electrical signal is input into a nonlinear baseline fitting model trained on cinema power supply data. This model, based on deep learning, constructs a dynamic baseline function and iteratively fits the nonlinear baseline trend of the power supply circuit as it changes with time, load, and temperature, bringing the signal baseline back to a stable reference level. After calibration, an artificial intelligence algorithm performs fine phase alignment on the disturbance and compensation components in the mixed electrical signal. Cross-correlation calculations are used to calculate the phase delay between the two components during transmission and coupling. Combined with the marked feature information, the timing position of the compensation signal is dynamically adjusted. Phase offset is corrected through iterative optimization to ensure that the peak and trough values ​​of the disturbance and compensation signals correspond synchronously, balancing amplitude consistency and waveform similarity. Errors are calculated in real-time using a loss function, and model parameters are adjusted until the two components are highly matched, outputting an aligned signal.

[0088] S44: The alignment processing signal is introduced into the grounding loop conduction suppression stage, and the harmonic blocking model is used to suppress the propagation of harmonics in the grounding loop, and the harmonic-limited power supply signal is output.

[0089] The aligned signal is then connected to the grounding loop conduction suppression stage. The AI-powered harmonic blocking model in this stage first analyzes the cinema grounding system topology, cable impedance distribution, and inter-device coupling characteristics to locate easily propagating harmonic nodes and high-frequency resonant points. Based on the marked harmonic characteristics, a dynamic suppression strategy is constructed. Targeted blocking is applied to excessive harmonic components, using phase reversal and amplitude clamping to prevent harmonics from propagating along the ground wire across devices. The suppression intensity is strictly controlled during the suppression process, only attenuating and isolating excessive harmonics. The AI ​​model monitors the grounding loop impedance and harmonic amplitude changes in real time, dynamically adjusting suppression parameters to adapt to load fluctuations under different operating conditions, ensuring stable suppression effects. After processing, the propagation of harmonics in the grounding loop is limited, high-frequency signal distortion and spike interference are significantly reduced, and the output harmonic-constrained power supply signal is produced.

[0090] S45: After optimizing the waveform and correcting the amplitude of the harmonic-limited power supply signal, verifying the cancellation effect and power supply stability, output the optimized power supply signal required by the cinema projection equipment.

[0091] During operation, AI-powered waveform optimization and amplitude correction are initiated. A deep learning waveform reconstruction algorithm precisely corrects residual distortion, signal dips, and abnormal spikes in the harmonic-constrained power supply signal, bringing the waveform closer to a standard sine wave. Combined with the rated voltage and current range of the projection equipment, closed-loop amplitude correction is performed, monitoring signal amplitude changes in real time and dynamically adjusting parameters to ensure voltage fluctuations do not exceed permissible deviations. After correction, the AI ​​evaluation module detects key indicators such as harmonic cancellation and total harmonic distortion (THD), comparing them with preset standards to determine compliance. If standards are not met, the model automatically backtracks to previous steps, adjusting parameters for feature extraction, fitting, and suppression, and re-executes the processing flow until all indicators are satisfactory, outputting an optimized power supply signal adapted to the cinema projection equipment.

[0092] S5: Based on the residual disturbance data of the optimized power supply signal, map it to the reconstructed phase space to calculate the Lyapunov exponential deviation, quantify the synchronization error between chaotic modulation and actual disturbance, and output closed-loop correction instructions.

[0093] S51: Collect and optimize residual disturbance data obtained from power supply signal retrieval, extract voltage amplitude and phase discrete values ​​point by point according to time sequence, remove abnormal sampling points and complete standardization processing, reconstruct phase points from one-dimensional time sequence array according to embedding dimension and delay time, map to high-dimensional space and generate phase space point set matrix.

[0094] Residual disturbance data remaining after optimized power supply signal compensation is collected from the signal re-acquisition of the cinema projection equipment's power supply circuit. The sampling frequency strictly matches the sampling frequency of the equipment's power supply signal. The core components of the re-acquisition signal are disassembled point-by-point according to time sequence, and the discrete values ​​of voltage amplitude and phase at each sampling moment are extracted. These two types of values ​​are arranged one-to-one according to time index, forming an original time series array containing timestamps, amplitudes, and phases. The Laida criterion is used to traverse the original time series array, calculating the array mean and standard deviation. Sampling points deviating from the mean by ±3 times the standard deviation are identified as outliers and replaced with the arithmetic mean of two adjacent valid sampling points. Zero-mean standardization is performed on the cleaned amplitude and phase value sequences to obtain a one-dimensional time series array. The mutual information value of the time series data is calculated using the mutual information method to determine the optimal delay time. The embedding dimension is gradually increased using the false nearest neighbor method until the proportion of false nearest neighbors is lower than a set threshold. Data segments with a length equal to the embedding dimension and a step size equal to the delay time are sequentially extracted from a one-dimensional time series array. Each data segment is mapped to a phase point in a high-dimensional space. All phase points are combined and arranged in the order of time evolution to generate a phase space point set matrix.

[0095] S52: Calculate the local expansion rate of the trajectory based on the phase space point set matrix, traverse the entire time series to obtain the Lyapunov exponent numerical sequence, and perform point-by-point difference operation between the Lyapunov exponent numerical sequence and the preset chaotic modulation reference exponent to obtain the Lyapunov exponent deviation numerical sequence.

[0096] For each phase point in the phase space point set matrix, the Euclidean distance formula is used to calculate its distance to all other phase points. A fixed number of neighboring phase points with the smallest distance are selected to form the neighborhood set of that phase point. The Euclidean distance between the central phase point and each neighboring phase point at the current time, and the Euclidean distance between them after a fixed time step, are recorded. The local expansion rate of the trajectory corresponding to the central phase point is obtained by taking the natural logarithm of the ratio of the two distances and dividing it by the time step. All time-series phase points in the phase space point set matrix are traversed in chronological order, and the above operations of neighborhood selection, distance calculation, and rate solving are repeated. The local expansion rates of each phase point are concatenated sequentially according to their time indices to obtain the initial local expansion rate sequence.

[0097] A pre-defined chaotic modulation baseline exponent sequence is retrieved. This sequence has the same time length, time step, and index correspondence as the calculated Lyapunov exponent sequence. At each time step, the two sets of exponent values ​​are subtracted, i.e., the difference between the Lyapunov exponent value and the corresponding baseline exponent value at each time step is calculated. The difference results are filtered to remove extreme values ​​caused by computational overflow or abnormal neighborhood selection. Linear interpolation is used to fill the gaps left after removing outliers. The remaining difference data are rearranged in chronological order to remove discontinuities caused by numerical jumps, ensuring continuous and smooth difference changes, thus forming a Lyapunov exponent deviation numerical sequence.

[0098] S53: Perform weighted statistics on the Lyapunov exponent deviation numerical sequence and calculate the mean square deviation value. Compare the deviation result with the preset threshold and output the synchronization error value.

[0099] Dynamic weighting coefficients are assigned to the Lyapunov index deviation numerical sequence in chronological order. The weighting coefficients are set in a linear increment, with higher weighting coefficients assigned to deviation data closer to the current time and lower weighting coefficients assigned to historical deviation data farther from the current time. The sum of the weighting coefficients is normalized to 1.

[0100] Each deviation value in the sequence is multiplied by its corresponding weight coefficient, and all products are summed to complete the weighted statistical processing of the deviation value sequence. The mean square deviation (MSD) value is calculated for the weighted deviation value sequence. First, the square of each weighted deviation value is calculated, then the arithmetic mean of all squares is calculated, and the square root of this mean is taken to obtain the MSD value that characterizes the overall deviation level. Preset multi-level numerical thresholds are retrieved, set according to the allowable range of power disturbances in the cinema projection equipment. The calculated MSD value is compared with each level of threshold to accurately determine the threshold interval where the deviation value lies. No additional logical judgments or functional conversions are performed; interval positioning is completed solely through pure numerical comparison. The interval identifier of the deviation value is converted into the corresponding quantized value. Redundant intermediate data generated during the weighted statistical process is removed, and only the core MSD quantization result is retained. The result is then formatted as a single-precision floating-point number, and invalid decimal places at the end of the value are removed to output the synchronization error value.

[0101] S54: Based on the synchronization error value, convert it into amplitude and phase correction parameters through a linear mapping relationship, integrate the parameter information, and generate closed-loop correction instructions.

[0102] Based on the synchronization error value, two independent linear mapping functions are constructed, corresponding to the mapping calculations of the amplitude correction parameter and the phase correction parameter, respectively. The form of the mapping function is set as follows: Where x is the synchronization error value, y is the correction parameter, k is the mapping slope, and b is the mapping intercept. The values ​​of k and b are pre-calibrated according to the rated parameters and correction accuracy requirements of the power supply of the cinema projection equipment.

[0103] The synchronization error values ​​are substituted into two linear mapping functions to obtain voltage amplitude correction coefficients and phase correction coefficients. Numerical limiting operations are performed on the two types of correction coefficients. Reasonable value ranges for amplitude correction coefficients and phase correction coefficients are set. The ranges strictly match the parameter receiving range of the power controller of the cinema projection equipment. Correction coefficients that exceed the range are clipped to the range boundary values ​​to avoid numerical failure due to parameters exceeding the hardware execution range.

[0104] The amplitude correction coefficient and phase correction coefficient are integrated and concatenated into a parameter group according to their fixed data formats. Based on the communication protocol of the cinema projection equipment's power controller, the concatenated parameter group is encoded and converted from floating-point correction parameters to binary instruction segments. According to the controller's required instruction frame structure, check bits, timing identifiers, and parameter length identifiers are filled in to ensure the instruction frame conforms to communication specifications. The check bits, timing identifiers, and binary parameter segments are concatenated to form a complete instruction frame. No additional control fields, function identifiers, or redundant information are added; only the core instruction content corresponding to the correction parameters is retained. Parity checks are performed on the instruction frame. After confirming there are no encoding errors, a closed-loop correction instruction conforming to the hardware interaction format is output.

[0105] S6: Update the coupling coefficient and manifold constraint of the dual-domain phase synchronization mapping according to the closed-loop correction instruction, correct the chaotic compensation trajectory in real time, so that the dynamic characteristics of the power supply signal remain stable within the working range allowed by the projection equipment, and output a high-quality power supply signal.

[0106] S61: Perform discrete numerical analysis on the closed-loop correction instruction and complete the amplitude normalization operation, outputting the standardized instruction feature vector.

[0107] When performing discrete numerical analysis on the closed-loop correction command, the continuous correction command is split into a discrete time sequence according to a fixed sampling interval. The sequence is then decomposed into numerical components of different scales through singular spectral decomposition. High-frequency noise components caused by line interference are removed. The effective sequence after decomposition is normalized in amplitude. During the processing, point-by-point numerical comparison is used to ensure that the discretization accuracy and normalization error are controlled within the set range. The output is a standardized command feature vector whose dimension matches the original command sequence.

[0108] S62: Substitute the standardized instruction feature vector into the two-domain phase synchronization mapping iterative equation to complete the numerical update of the coupling coefficient and manifold constraint conditions, and output the mapping constraint parameter set.

[0109] When the standardized instruction feature vector is substituted into the iterative equation of the two-domain phase synchronization mapping, the values ​​of each dimension of the feature vector are used as the initial driving force for phase synchronization. An iterative update operator is constructed by combining the chaotic phase modulation rule. The numerical correlation between the feature vector and the original phase parameters of the mapping is calculated through cross-correlation operation. Based on the correlation result, the coupling coefficient is iteratively adjusted to keep the coefficient value dynamically matched with the instruction feature. The boundary values ​​and curvature parameters in the manifold constraint conditions are synchronously corrected. The parameter update range is constrained by the manifold embedding matching algorithm to avoid the constraint conditions from exceeding the effective numerical range of phase synchronization. The entire iterative process continuously updates the values ​​with the goal of minimizing the phase deviation until the parameters converge to a stable state. After all the numerical adjustments are completed, the mapping constraint parameter set containing the coupling coefficient and constraint parameters is output.

[0110] S63: Perform point-by-point numerical correction on the chaotic compensation trajectory based on the mapping constraint parameter set, and output the numerical sequence of the compensation trajectory.

[0111] When performing point-by-point numerical correction on the chaotic compensation trajectory based on the mapping constraint parameter set, the original coordinate sequence of the chaotic compensation trajectory is read. The coupling coefficient in the mapping constraint parameter set is used as the weighting factor for trajectory adjustment, and the manifold constraint condition is used as the boundary constraint condition for trajectory values. The trajectory coordinates are traversed point by point and nonlinear baseline fitting is performed. Drift adaptive correction is performed on trajectory nodes that exceed the constraint range. Redundant fluctuation values ​​in the trajectory are separated by the sparse Bayesian peak unmixing algorithm. During the correction process, the change rate of node values ​​before and after is compared in real time to avoid numerical jumps in the trajectory due to excessive single correction. The updated value is recorded after each node correction is completed, and the compensation trajectory numerical sequence is output.

[0112] S64: Perform real-time numerical calibration and amplitude limiting filtering on the dynamic characteristics of the power supply signal based on the numerical sequence of the compensation trajectory, and output the numerical sequence of the power supply signal.

[0113] When performing real-time numerical calibration and amplitude limiting filtering on the dynamic characteristics of the power supply signal based on the compensation trajectory numerical sequence, the difference between the compensation trajectory numerical sequence and the original dynamic characteristic values ​​of the power supply signal is calculated point by point. The characteristic values ​​are fine-tuned based on the difference results so that the values ​​gradually approach the working range allowed by the projection equipment. After calibration, amplitude limiting filtering is introduced, and upper and lower limit thresholds corresponding to the working range of the equipment are set. Characteristic values ​​exceeding the threshold are truncated. The numerical abrupt changes caused by amplitude limiting are smoothed by moving average filtering. During the filtering process, the fundamental frequency and amplitude characteristics of the power supply signal are preserved without changing its core dynamic structure. The calibration and filtering parameters are continuously and synchronously updated during the processing, and the output power supply signal numerical sequence with stable dynamic characteristics, no abnormal fluctuations, and fully adapted to the working requirements of cinema projection equipment is obtained.

[0114] like Figure 3As shown, the present invention also provides an artificial intelligence-based power management system for cinema projection equipment, comprising: The signal reconstruction module is used to acquire power supply harmonic current and zero-ground voltage fluctuation signals, process them through singular spectrum decomposition and phase space reconstruction, remove noise and normalize the timing, and output a standardized electrical signal sequence.

[0115] The phase domain solution module is used to input standard electrical signals into the dual-domain phase synchronization mapping model, solve the manifold offset based on the characteristics of the chaotic system, and generate a disturbance feature vector by fusing dual-domain features.

[0116] The phase prediction module is used to construct a sparse Bayesian model based on the perturbation feature vector, predict the harmonic amplitude and phase changes, and drive the chaotic oscillator to generate a leading phase compensation electrical signal.

[0117] The phase cancellation suppression module is used to inject the compensation signal into the power supply circuit, and achieve precise phase cancellation of disturbances through nonlinear baseline fitting, suppress the spread of harmonics in the grounding circuit, and output an optimized power supply signal.

[0118] The error quantization module is used to acquire residual disturbance data of the optimized signal, reconstruct the phase space and calculate the Lyapunov exponential deviation, quantify the synchronization error and generate closed-loop correction instructions.

[0119] The closed-loop correction module is used to update the mapping parameters and constraints according to the correction instructions, correct the chaotic compensation trajectory in real time, and output a high-quality power supply signal.

[0120] In summary, this embodiment provides an AI-based power management method and system for cinema projection equipment. By using closed-loop disturbance retrieval, Lyapunov exponential deviation quantization, and dual-domain phase synchronization mapping for real-time correction, a complete closed-loop control process from compensation, suppression, feedback, to instruction correction is constructed. The coupling coefficient and manifold constraints are continuously updated and the chaotic compensation trajectory is corrected. This achieves the beneficial effects of dynamically eliminating residual disturbances, maintaining stable power supply signal dynamics, and locking the power supply quality within the equipment's allowable operating range for a long period. It effectively suppresses harmonic propagation in the grounding loop, reduces the impact of power supply interference on projection quality and equipment lifespan, and achieves self-optimization and self-stabilization of power supply quality. It can adapt to long-term continuous projection scenarios without manual intervention, significantly improving the operational stability and intelligent level of power management of cinema projection equipment.

[0121] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0122] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power management method for cinema projection equipment based on artificial intelligence, characterized in that, include: S1: Acquire the harmonic current waveform and zero-ground voltage fluctuation signal of the power port of the cinema projection equipment, perform singular spectrum decomposition and phase space reconstruction on the original disturbance signal, and output a standard electrical signal sequence. S2: Input the standard electrical signal sequence into the dual-domain phase synchronization mapping model, and use the synchronization deviation sensitivity of the chaotic system to solve the unstable manifold offset of harmonics and zero-ground disturbance in phase space, and generate the disturbance feature vector. S3: Construct a sparse Bayesian phase prediction structure based on the disturbance feature vector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at future moments, drive the chaotic oscillator to generate a leading phase compensation waveform based on the prediction results, and output the compensation electrical signal. S4: The compensation electrical signal is injected into the front-end power circuit of the projection equipment through the coupling structure. The disturbance and compensation signal are precisely canceled by nonlinear baseline fitting and alignment, the diffusion and propagation of harmonics in the grounding circuit are suppressed, and the optimized power supply signal is output. S5: Based on the residual disturbance data of the optimized power supply signal, map it to the reconstructed phase space to calculate the Lyapunov exponential deviation, quantify the synchronization error between chaotic modulation and actual disturbance, and output closed-loop correction instructions; S6: Update the coupling coefficient and manifold constraint of the dual-domain phase synchronization mapping according to the closed-loop correction instruction, correct the chaotic compensation trajectory in real time, so that the dynamic characteristics of the power supply signal remain stable within the working range allowed by the projection equipment, and output a high-quality power supply signal.

2. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S1, the specific steps for outputting the standard electrical signal sequence are as follows: S11: Collect the harmonic current waveform and ground voltage fluctuation original disturbance signal of the power port of the cinema projection equipment, and perform equal-interval resampling and outlier removal, remove DC offset components, and construct one-dimensional original time series data. S12: Construct a phase space trajectory matrix from the one-dimensional original time series data according to a preset embedding dimension and delay time. Separate the effective components and random noise components that characterize the disturbance features through orthogonal decomposition of the trajectory matrix to obtain the core disturbance information. S13: The core disturbance information is sorted and filtered by singular values, the main singular value components corresponding to the power disturbance are extracted, and the random interference corresponding to the weak singular values ​​is removed to obtain the reconstructed disturbance signal. S14: Perform amplitude normalization processing on the reconstructed disturbance characteristic signal to eliminate dimensional differences and amplitude deviations, and arrange it into segments according to a fixed time window to form a standard electrical signal sequence.

3. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S2, the specific steps for generating the perturbation feature vector are as follows: S21: Based on the power supply disturbance characteristics of cinema projection equipment, the standard electrical signal sequence is split into time-domain zero-ground voltage fluctuation and frequency-domain harmonic distribution, and the phase point trajectory and frequency domain characteristics that fit the load of the laser projector are constructed to form a dual-domain initial feature set. S22: Substitute the initial features of the dual domain into the chaotic dynamic equation that adapts to the perturbation bandwidth of the projection equipment for iterative mapping, and combine the instantaneous power impact characteristics to align the features with the chaotic attractor to obtain the phase space distribution results. S23: Based on the synchronous deviation response of the chaotic system to the harmonics of the projection load and the zero-ground coupling disturbance, the displacement data of the trajectory deviating from the reference attractor is extracted from the phase space distribution, and the unstable manifold offset of the projection equipment disturbance is calculated. S24: The unstable manifold offset and the dual-domain feature are normalized and concatenated according to the projection equipment disturbance weight, arranged in a fixed dimension as a one-dimensional continuous vector, and the disturbance feature vector for cinema projection power supply phase compensation is output.

4. The power management method for cinema projection equipment based on artificial intelligence according to claim 3, characterized in that: In step S23, the specific steps for obtaining the unstable manifold offset are as follows: Locate the center of the ideal attractor of the chaotic system in the phase space distribution results, construct a local coordinate reference system, calculate the Euclidean distance between each phase point and the attractor center point by point, and form an initial displacement distribution sequence. The initial displacement distribution sequence is subjected to nonlinear smoothing to remove abrupt displacement values ​​and retain the trend displacement components that are coupled with harmonics and zero-ground disturbances, thus obtaining the displacement change sequence. Based on the displacement change sequence, tangential and normal decompositions are performed along the phase space trajectory to solve for the displacement extremes and average offset rates in the two directions, thereby obtaining a set of key offset parameters. Substituting the set of key offset parameters into the manifold curvature calculation, the positional deviation and angular deflection of the disturbance on the unstable manifold are obtained, thus yielding the unstable manifold offset corresponding to the power supply disturbance of the projection equipment.

5. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S3, the specific steps for outputting the compensation electrical signal are as follows: S31: Based on the perturbation feature vector, the prior distribution of the sparse Bayesian model is initialized to construct a basis function space that adapts to the harmonic phase change. The feature dimension is reduced by automatic correlation determination to obtain the sparse Bayesian prediction model. S32: Based on the sparse Bayesian prediction model and the historical dynamic laws in the perturbation feature vector, perform nonlinear recursive estimation of the harmonic amplitude and phase changes at multiple future moments, and output the harmonic phase prediction sequence. S33: Based on the harmonic phase prediction sequence and the amplitude fluctuation parameter in the disturbance feature vector, tune the iterative parameters and initial state of the chaotic oscillator to generate a chaotic compensation waveform sequence; S34: Perform amplitude normalization and timing alignment calibration on the chaotic compensation waveform sequence to form a compensation electrical signal that perfectly matches the power supply disturbance of the cinema projection equipment.

6. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S4, the specific steps for outputting the optimized power supply signal are as follows: S41: Collects the original compensation electrical signal of the power supply circuit at the front end of the cinema projection equipment, analyzes the harmonic components and noise distribution of the signal through an artificial intelligence time-series feature extraction algorithm, and outputs the feature-marked compensation electrical signal; S42: The feature marker compensation electrical signal is injected into the front-end power circuit of the projection equipment through a coupling structure, the real-time disturbance waveform of the circuit is collected and the signal is superimposed, and a mixed electrical signal is output. S43: The hybrid electrical signal is baseline-calibrated and phase-aligned using an artificial intelligence nonlinear baseline fitting algorithm, and an alignment-processed signal is output. S44: The alignment processing signal is introduced into the grounding loop conduction suppression stage. The harmonic blocking model is used to suppress the propagation and conduction of harmonics in the grounding loop, and the harmonic-limited power supply signal is output. S45: After optimizing the waveform and correcting the amplitude of the harmonic-limited power supply signal, verifying the cancellation effect and power supply stability, output the optimized power supply signal required by the cinema projection equipment.

7. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S5, the specific steps for generating the closed-loop correction command are as follows: S51: Collect and optimize the residual disturbance data obtained from the power supply signal retrieval, extract the voltage amplitude and phase discrete values ​​point by point according to the time sequence, remove abnormal sampling points and complete the standardization process, reconstruct the phase points of the one-dimensional time sequence array according to the embedding dimension and delay time, map it to the high-dimensional space and generate the phase space point set matrix. S52: Calculate the local expansion rate of the trajectory based on the phase space point set matrix, traverse the entire time series to obtain the Lyapunov exponent numerical sequence, and perform point-by-point difference operation between the Lyapunov exponent numerical sequence and the preset chaotic modulation reference exponent to obtain the Lyapunov exponent deviation numerical sequence. S53: Perform weighted statistics on the Lyapunov exponent deviation value sequence and calculate the mean square deviation value. Compare the deviation result with a preset threshold and output the synchronization error value. S54: Based on the synchronization error value, convert it into amplitude and phase correction parameters through a linear mapping relationship, integrate the parameter information, and generate a closed-loop correction command.

8. The power management method for cinema projection equipment based on artificial intelligence according to claim 7, characterized in that: In step S52, the specific steps for obtaining the numerical sequence of Lyapunov exponential deviations are as follows: Based on the phase space point set matrix, traverse each phase point in the matrix and select neighboring phase points, calculate the Euclidean distance and time evolution difference between neighboring phase points, and output the local expansion rate array corresponding to each phase point; The local expansion rate array is subjected to time-series weighted fitting, and the average exponential growth rate is calculated segment by segment along the trajectory evolution direction. The calculation results of the whole time period are integrated and normalized to output the Lyapunov exponential numerical sequence. The Lyapunov exponent numerical sequence and the preset chaotic modulation benchmark exponent are calculated point by point according to the same time index. The difference result at each time point is recorded and the original difference sequence is output. The original difference sequence is subjected to outlier removal and smoothing. After removing discrete jump points, it is rearranged according to time sequence to output the Lyapunov exponential deviation numerical sequence.

9. The power management method for cinema projection equipment based on artificial intelligence according to claim 1, characterized in that: In step S6, the specific steps for outputting a high-quality power supply signal are as follows: S61: Perform discrete numerical analysis on the closed-loop correction instruction and complete the amplitude normalization operation to output the standardized instruction feature vector; S62: Substitute the standardized instruction feature vector into the two-domain phase synchronization mapping iterative equation to complete the numerical update of the coupling coefficient and manifold constraint conditions, and output the mapping constraint parameter set; S63: Perform point-by-point numerical correction on the chaotic compensation trajectory according to the mapping constraint parameter set, and output the numerical sequence of the compensation trajectory. S64: Perform real-time numerical calibration and amplitude limiting filtering on the dynamic characteristics of the power supply signal based on the compensation trajectory numerical sequence, and output the power supply signal numerical sequence.

10. An AI-based power management system for cinema projection equipment, comprising the AI-based power management method for cinema projection equipment as described in any one of claims 1 to 9, characterized in that, The management system includes: The signal reconstruction module is used to acquire power supply harmonic current and zero-ground voltage fluctuation signals, and output a standardized electrical signal sequence after singular spectrum decomposition and phase space reconstruction. The phase domain solution module is used to input the standard electrical signal into the dual-domain phase synchronization mapping model, solve the manifold offset according to the characteristics of the chaotic system, and generate the disturbance feature vector by fusing the dual-domain features. The phase prediction module is used to construct a sparse Bayesian model based on the disturbance feature vector, predict the harmonic amplitude and phase changes, and drive the chaotic oscillator to generate a compensation electrical signal. The phase cancellation suppression module is used to inject the compensation signal into the power supply circuit, perform precise phase cancellation of disturbances through nonlinear baseline fitting, and output an optimized power supply signal; The error quantization module is used to collect residual disturbance data of the optimized power supply signal, reconstruct the phase space and calculate the Lyapunov exponential deviation, quantify the synchronization error and generate closed-loop correction instructions. The closed-loop correction module is used to update the mapping parameters and constraints according to the correction instructions, correct the chaotic compensation trajectory in real time, and output a high-quality power supply signal.

Images