106 results about "Noise optimization" patented technology

The invention discloses a big data privacy protection modeling method and system based on federated learning and a block chain, and relates to the technical field of privacy protection and joint modeling. According to the method, homomorphic encryption, differential privacy, federated learning, secure multi-party computing and block chain technologies are fused, big data privacy protection and joint modeling are realized, encryption and dimensionality reduction are performed on original data through homomorphic encryption and differential privacy, an encrypted training sample of secure privacy is generated, a local model is trained on an encrypted data set through federated learning, and a big data privacy protection result is obtained. The method comprises the following steps: calculating aggregation parameters by using security multiple parties, constructing a verification network in combination with a block chain, ensuring credibility and integrity of model training, and finally, adding noise optimization performance for a global model by using differential privacy, testing generalization ability through cross validation, and determining a deployable privacy protection joint learning model, thereby breaking traditional data islands, promoting cross-mechanism data cooperation, and improving the privacy protection performance. Big data values are released, and data protection regulations and privacy requirements are met.

The invention relates to the technical field of vibration noise optimization, in particular to a vibration noise optimization method and system of an electric drive assembly. The method comprises the following steps of collecting multi-working-condition vibration noise of a transmission motor in an electric drive assembly and performing frequency domain conversion to generate vibration noise frequency domain data; on the basis of the data, carrying out gear meshing stress unbalance loading limit calculation, carrying out tooth surface pitting corrosion cycle ductility simulation, and identifying a bearing clearance expansion increment; optimizing the bearing structure according to the bearing clearance expansion incremental data to obtain bearing structure optimization data; performing tooth direction drum shape modification analysis according to the tooth surface pitting corrosion data to obtain modification data; and finally, carrying out anti-vibration strengthening design by combining bearing optimization and tooth direction modification data, generating structural anti-vibration strengthening design data, and sending the structural anti-vibration strengthening design data to a terminal so as to execute structural optimization and reduce the influence of vibration noise. According to the method, the vibration noise optimization technology is optimized, so that the vibration noise optimization technology is more perfect.

The invention relates to a programmable frequency divider applied to JESD204B / C. The programmable frequency divider comprises a prescaler which realizes 1 / 2 / 4 dynamic frequency division through a closed-loop feedback structure; according to the 12-bit main frequency divider, the duty ratio of odd-even frequency division is controlled by a 11-stage asynchronous CML frequency-halving chain and a separation DFF0 module; the frequency dividing ratio N is equal to DIV Nlt; 11: 0gt; 2, the odd number frequency division duty ratio is automatically adjusted to (N + 1) / 2N; the cascade noise optimization comprises the following steps of: resampling by using an input clock fin through a retiming trigger (RetimeDFF1 / 2); according to the pulse generator, the number K of pulses output by a four-stage counter is equal to Pulsecount; 3: 0gt; a decimal value; and the output mode circuit supports three-mode direct connection. According to the programmable frequency divider, the HBT process CML unit and the emitter follower are adopted for buffering; and the high-precision clock synchronization requirements of JESD204B / C systems such as a 5G base station and a millimeter wave radar can be met.

The invention relates to the technical field of reducer vibration noise optimization, and discloses a planet row coaxial reducer vibration noise optimization method, which comprises the steps of obtaining initial assembly parameters of a reducer, collecting load measurement signals of planet wheels of the reducer under an order frequency sweeping working condition, converting the load measurement signals into equivalent meshing loads based on a calibration curve, and calculating the equivalent meshing loads according to the equivalent meshing loads. Calculating a load imbalance coefficient according to the equivalent meshing load, and taking a peak value of the load imbalance coefficient as a first evaluation index; shell radial acceleration and sound pressure signals are synchronously collected under the order frequency sweeping working condition, the side-band ratio is calculated, and the peak value of the side-band ratio is taken as a second evaluation index; and performing diagnosis shunting based on the first evaluation index and the second evaluation index, determining an assembly parameter which is adjusted preferentially, executing iterative adjustment, and outputting a final assembly parameter after iteration is completed. According to the method, uniform load unbalance caused by assembly errors and side band amplification caused by geometric eccentricity can be effectively separated and solved, and vibration and noise of the speed reducer are reduced fundamentally.

The invention discloses an intelligent obstacle recognition method for an unmanned loader, and aims to solve the problems of low obstacle recognition precision, poor real-time performance and insufficient coordination of obstacle avoidance and operation in a complex operation scene. The method comprises the following steps: establishing a multi-source sensing system such as a laser radar and a camera, and realizing data time sequence synchronization, space unification and noise optimization through FPGA hardware; obstacle space and category features are extracted through double-branch parallel detection, and a modal weight fusion result is dynamically distributed in combination with a scene; an improved Hungary algorithm and a classification tracking model are adopted to realize obstacle ID stable association and trajectory tracking, and trajectory prediction is completed based on LSTM and Kalman filtering; and a dynamic avoidance area is constructed, and an obstacle avoidance path is generated through global-local bilevel programming and a safety potential field algorithm and is corrected in real time. The obstacle recognition robustness and real-time performance in a complex environment are improved, and safe and efficient operation of the unmanned loader is guaranteed.

The invention discloses an unmanned aerial vehicle propeller design based on aerodynamic performance and noise optimization, and relates to the field of propeller structure design. Aiming at the problem of contradiction between performance and noise optimization design during design of an existing propeller structure, a new design and research thought of the propeller structure for the unmanned aerial vehicle is given, and the propeller structure for the unmanned aerial vehicle which meets aerodynamic performance and is low in aerodynamic noise is designed in a multi-main-microparameter cooperation mode. According to the research thought, the influence effect on the aerodynamic performance and noise of the propeller under different measures can be analyzed, meanwhile, the aerodynamic performance and noise reduction are considered, the design scheme of the propeller for the unmanned aerial vehicle with the maximum take-off thrust, the optimal cruise efficiency and the minimum operation noise is designed under the condition that the service time and the service life of the propeller are guaranteed, and the design method is suitable for popularization and application. The thinking steps are simple and short in time consumption, and basis is provided for subsequent propeller design in the same field.

The invention discloses an image scene conversion method and device, equipment and a storage medium. Firstly, an image to be processed and a conversion cue word used for indicating an image scene conversion direction and a specific attribute adjustment requirement can be obtained, the conversion cue word is input into a pre-trained scene noise optimizer, and structured guide noise accurately matched with the cue word is generated through noise optimization processing and used as target noise. Performing vector encoding on the to-be-processed image and the conversion prompt word through an image encoder and a text encoder to obtain an image feature vector and a text feature vector, and performing feature fusion on the image feature vector and the text feature vector by using a cross attention mechanism to generate a fusion vector; and finally, jointly inputting the fusion vector and the target noise into a pre-trained scene conversion model, and performing conditional decoding generation processing to obtain a scene conversion image corresponding to the to-be-processed image. According to the invention, disordered disturbance caused by noise injection is avoided, and drift of the image content in the conversion process is effectively inhibited.

The invention discloses a high-accuracy house type image inner contour segmentation method, system and device. The method comprises the following steps: carrying out normalization processing (such as data enhancement and noise optimization) on data; detecting a target area based on an improved DETR model, extracting multi-layer features through an optimized encoder and decoder, and generating a bounding box and confidence and category thereof; combining the bounding box with the original image, randomly expanding the bounding box to extract a house area, and reducing watermark and advertisement interference in the image; and on the basis of an improved Mask2Former model, an optimized Transform decoder and an attention mechanism are utilized to generate a segmentation mask and a classification result. According to the method, DETR and Mask2Former models are combined, all functional rooms of the house type image are quickly and accurately segmented, and the blank of a large-scale house type image processing method is filled.

The invention provides an image generation method and device, equipment and a storage medium, and the method is applied to the field of computer images, and comprises the steps: obtaining a style reference image of a to-be-processed image if the style guide type of the to-be-processed image is an image guide type, determining a target image style corresponding to the style reference image based on the content feature and the style feature of the style reference image; if the style guide type is an instruction guide type, obtaining a guide text instruction of the to-be-processed image, and determining a target image style corresponding to the guide text instruction based on text feature information of the guide text instruction; adding random noise to the to-be-processed image to generate a noise image; and generating a noise optimization parameter for the noise image based on the target image style, and calling the noise optimization parameter to carry out iterative denoising on the noise image to generate a target style image. Therefore, the precision and efficiency of style migration are improved, and the detail fidelity and style consistency of the target style image are enhanced at the same time.

The invention discloses an audio noise reduction method, system and device based on an AI model and a storage medium, and relates to the technical field of voice noise reduction, and the method comprises the steps: receiving a live video stream, and segmenting the live video stream into image frame data and audio stream data; the image frame data and the audio stream data are input to a preset scene recognition model, the recognition result of the current scene is obtained, and the preset scene recognition model comprises a character recognition sub-model, a voice recognition sub-model and a classification output module; and determining and executing a current noise reduction strategy according to the identification result to obtain an audio output signal, the current noise reduction strategy being a human voice enhancement strategy, an ambient sound enhancement strategy or a fusion enhancement strategy. According to the invention, scene recognition is carried out in a multi-modal fusion mode, different noise reduction strategies are adaptively executed according to the scene recognition result, noise optimization can be carried out for different scenes, and the audio output effect is improved.

The invention provides a method for testing noise and vibration of a refrigeration fan of a refrigerator, relates to the field of vibration testing of refrigerators, and aims to solve the problem that the real noise performance of the fan in a complex environment in the refrigerator cannot be reflected in the current noise testing of the fan of the air-cooled refrigerator. The actual installation states of a fan, an air duct and a box body are reserved to eliminate single body test deviation, vibration and noise data are synchronously collected, time domain and frequency domain analysis is covered, vibration-noise causal relation tracing is achieved, the coupling frequency band and energy distribution of vibration and noise are quantified through time domain and frequency domain conjoint analysis, and therefore the noise source and the coupling frequency band are positioned, and the noise quality is improved. And targeted optimization is facilitated, and the noise optimization efficiency and the optimized noise control effect are improved.

The invention discloses a multi-source data fusion method and system based on data analysis, and relates to the technical field of electric digital data processing. The multi-source data fusion method based on data analysis comprises the following steps of performing fusion real-time analysis; performing dynamic fusion adaptive analysis; and interactive fusion validity analysis. According to the method, real-time analysis is carried out through real-time fusion data, whether sampling sensing frequency optimization is carried out or not is judged, then dynamic adaptability analysis is carried out according to sensing fusion data, whether sensing fusion parameter optimization is carried out or not is judged, finally interaction effectiveness analysis is carried out, and whether interaction fusion noise optimization is carried out or not is judged. Therefore, the reliability of the corresponding fire-fighting information dynamic fusion of the fire-fighting facilities in the transformer substation in the communication process of the internet of things is improved, and the problem that the reliability of the corresponding fire-fighting information dynamic fusion of the fire-fighting facilities in the transformer substation in the communication process of the internet of things is not high in the prior art is solved.

The invention relates to the technical field of voice recognition, in particular to an electric power self-service terminal voice intention recognition system based on background noise optimization. The system comprises a voice interaction noise optimization module, a scene voice transcription analysis module, a scene voice intention recognition module and a recognition confidence iterative optimization module, a voice interaction original data set during voice interaction between the power self-service terminal and a user can be collected, and noise optimization is carried out to generate a target voice interaction data set; performing voice transcription analysis based on the target voice interaction data set, and generating a text transcription result; performing keyword semantic matching calculation on the text transcription result and an intention recognition knowledge base to obtain an intention recognition result of the user; and calculating a comprehensive recognition confidence coefficient based on the intention recognition result, if the comprehensive recognition confidence coefficient is lower than a set threshold value, performing secondary noise optimization on the voice interaction original data set and re-executing the judgment step, and if not, outputting the intention recognition result of the user. According to the invention, the accuracy of terminal speech recognition and intention recognition can be improved.

The invention belongs to the technical field of new energy automobile noise optimization, and particularly relates to a gear modification method for reducing gear noise, which comprises the following steps: S1, constructing a gear transmission system model; s2, simulating gear transmission performance under different load conditions and extracting key parameters; s3, shaping parameters to be optimized are set; s4, performing optimization calculation on each to-be-optimized modification parameter by taking reduction of a transmission error and improvement of a gear contact area as targets; in the optimization calculation process, according to a preset weight distribution strategy, weight proportions are set for different operation conditions and used for representing priorities of all the operation conditions; and S5, adjusting the numerical value of the modification parameter in the optimization result in the step S4, so that the transmission error curve is in a sine or cosine trend under all operation conditions. According to the method, smoothing of a transmission error curve, optimization of tooth surface contact performance and uniformization of edge load distribution can be achieved within the full working condition range, gear meshing noise is remarkably reduced, and the noise quality of a whole vehicle is improved.

The application relates to a noise optimization control method, device and equipment for rotor capture and a storage medium, and relates to the fan technical field, and particularly relates to a noise optimization control method, device and equipment for rotor capture and a storage medium, which comprises the following steps: obtaining a starting instruction issued by a fan; outputting first voltage data through voltage detection and comparison by using the starting instruction issued by the fan; determining the motion state of the fan at present by using the first voltage data; and obtaining the related results of noise optimization control of single-resistance sampling during rotor capture by using the determination result of the motion state of the fan at present and executing the corresponding motion mode. Through voltage detection and comparison, the first voltage data is outputted, the motion state of the fan at present is determined by using the first voltage data, and the corresponding motion mode is executed, so that single-resistance sampling realizes noiseless starting during rotor capture.

The invention discloses a city design noise optimization method and system based on artificial intelligence, and the method comprises the steps: carrying out the preprocessing of building form and noise measurement data, and constructing a basic data set after noise simulation verification; a three-dimensional city model and a measuring point network are generated through a layered parameterization framework, the structural noise level is calculated, real-time verification is carried out in combination with planning constraints, and a structural design space, noise basic data and feasibility guarantee are provided; a two-stage multi-layer perceptron network is constructed, local building group acoustic influence and a global city noise mode are processed respectively, and high-precision noise prediction is achieved through training; and deploying a large language model as an optimization agent, and outputting an optimal scheme through multiple rounds of iterative optimization and evaluation. Compared with the prior art, the method has the advantages that the noise prediction precision is effectively improved, a structured design space is provided for parametric modeling, the feasibility is guaranteed through real-time verification, efficient iteration is achieved through large language model optimization, and compared with a traditional method, the efficiency is remarkably improved, and the labor cost is reduced.

The present invention discloses a traffic classification method and system based on smooth noise data enhancement. The method comprises the following steps: a noise generation module generates unified normalized noise and converts the normalized noise into characteristic noise; a noise optimization module optimizes the noise shape through a gradient-based algorithm and optimizes the noise scale through an improved Bayesian optimizer; a data enhancement module generates optimized smooth noise based on the optimized noise shape and noise scale, and adds the optimized smooth noise to benign and malicious samples, respectively, to obtain enhanced new samples; an integrated learning module uses the enhanced samples as pseudo-malicious samples to train a traffic classifier, collects the results of traffic classifiers trained under multiple sets of noise scales, and combines the classification results using an improved Boosting algorithm to obtain the final classification result. The method effectively optimizes the classification boundary of the traffic classifier, improves the recall rate of the traffic classifier for unknown types of malicious traffic, and enhances the generalization of the classification results.

The invention relates to a ducted propeller (1) and to an aircraft having a ducted propeller according to the invention. The ducted propeller (1) has a stator (10) which comprises a central element (12) extending along a longitudinal axis (x). The stator (10) further has a housing (13) which is arranged around the central element (12), wherein the central element (12) is connected to the housing (13) via at least one first web (11a) and at least one second web (11b). A rotor (20) having a plurality of rotor blades (21) is arranged on the central element (12) of the stator (10).According to the invention, a first web (11a) is arranged offset from a second web (11b) axially along the longitudinal axis (x) by a defined distance (d), wherein the distance (d) is selected such that a first sound wave (4a) generated at the first web (11a) due to the air flow is 180° out of phase with a second sound wave (4b) generated at the second web (11b) due to the air flow and identical to the first sound wave (4a).

The present invention discloses a variable-leakage permanent magnet synchronous motor and its low-vibration-noise optimization design method that considers multiple operating conditions. The method comprises the following steps: obtaining a variable permanent magnet leakage flux through the magnetic circuit distribution of the permanent magnet motor; adjusting the amount of permanent magnet leakage flux to meet the motor's requirements for torque, speed regulation range, and low vibration and noise when the motor is in different operating conditions; constructing a regional leakage flux structure to achieve controllable leakage flux and provide the motor's initial structure; analyzing the electromagnetic vibration characteristics of the variable-leakage permanent magnet synchronous motor under different operating conditions to obtain relevant data on performance such as the lowest non-zero-order electromagnetic force; evaluating the degree to which key motor parameters affect electromagnetic performance and the performance requirements of the motor under different operating conditions; and performing comprehensive optimization using a multi-objective genetic algorithm to determine the final size of the motor. The present invention enables the motor to achieve wide speed regulation and high torque performance while suppressing the motor's vibration and noise under different operating conditions.

The present invention is applicable to the field of knowledge graph technology and provides an entity alignment method based on noise optimization and attribute fusion, comprising the following steps: performing noise initialization on an entity seed set to obtain accurate noise entity pairs; injecting a noise-aware optimization generator to obtain more noise entity pairs around the relevant distribution of true seed entity pairs; after generating noise entity pairs, feeding them into a noise-aware optimization discriminator to distinguish true seed entity pairs from noise entity pairs; iterating the noise-aware optimization generator and the noise-aware optimization discriminator to obtain a trust score, which is compared with a threshold θ to obtain clean labeled data; structurally embedding the obtained true seed entity pairs through GCN to ultimately obtain a structural alignment model. This method utilizes a combination of noise optimization and attributes to not only reduce complexity but also improve the effect of entity alignment, achieving excellent entity alignment results while reducing experimental overhead.

The invention relates to the technical field of vehicles, in particular to an electric vehicle noise optimization method and device, a vehicle, a storage medium and a product, and the method comprises the steps that an in-vehicle microphone is used for collecting in-vehicle noise data of the vehicle under the current working condition; extracting frequency spectrum characteristics of the in-vehicle noise data, removing random noise in the in-vehicle noise data according to the frequency spectrum characteristics, and determining target playing parameters of an in-vehicle loudspeaker according to the frequency spectrum characteristics of the in-vehicle noise data after the random noise is removed; and controlling an in-vehicle loudspeaker to play audio according to the target playing parameter to optimize the in-vehicle noise. Therefore, the problems that the effect is limited, the cost is too high or extra noise is introduced, and high-efficiency, low-cost and practical noise reduction is difficult to realize when the high-frequency noise of the motor of the electric vehicle is treated in the related technology are solved.

The invention relates to a combined type energy storage equipment noise reduction system and a noise optimization method. The combined type energy storage equipment noise reduction system comprises a noise reduction cover body arranged corresponding to a refrigeration assembly, and a noise reduction control module is further arranged in the noise reduction cover body; the noise reduction cover body comprises a perforated aluminum plate and a split type sound absorption plate, the perforated aluminum plate is used as a high-frequency noise reflecting layer, the split type sound absorption plate is used as a medium-frequency noise dissipation layer, the noise reduction cover body further comprises a bottom vibration isolation module, and the bottom vibration isolation module is used as a low-frequency vibration isolation layer for restraining low-frequency vibration transmission; and combining to form a physical filtering noise reduction structure for suppressing broadband noise through a gradient structure.

The invention relates to the technical field of communication, in particular to an ultra-wideband low-noise amplifier based on a coupling expansion technology. The ultra-wideband low-noise amplifier comprises a Cascode amplifying circuit, an input coupling inductor pair and an output coupling inductor pair. A drain stray capacitor Cd and a load capacitor CL in a circuit are rapidly charged through negative electromotive force generated by the mutual inductance effect, the charging time is remarkably shortened, the working bandwidth of the amplifier is greatly expanded on the premise that the noise performance is not sacrificed, meanwhile, the requirement for a multi-stage impedance matching network is simplified through the inherent broadband characteristic of the coupling structure, and the cost is reduced. According to the ultra-wideband power divider, ultra-wideband input and output matching is achieved, noise matching is optimized through the noise optimization capacitor Cex, the chip area and the number of peripheral elements are effectively reduced through the multiplexing inductor function, and finally the contradiction among bandwidth, noise, matching and area power consumption is cooperatively solved under the condition of high integration level.

The invention relates to an idling charging noise optimization method and device of a vehicle, electronic equipment and a medium, and the method comprises the steps: calculating the suspension loading force of the vehicle through measuring a related transfer function, the actual vibration acceleration of an indication point and the actual sound pressure of a target response point under the idling charging working condition of the vehicle by using an inverse matrix method; coupling the whole vehicle model and the acoustic cavity finite element model of the vehicle to obtain a whole vehicle interior noise vibration prediction simulation model; inputting the suspension loading force into the whole vehicle interior noise vibration prediction simulation model, and predicting the vehicle interior vibration noise under the idling charging excitation of the whole vehicle state of the vehicle; and identifying a problem characteristic frequency meeting a preset optimization condition from the in-vehicle vibration noise, and determining an idling charging noise optimization scheme of the vehicle based on the problem characteristic frequency so as to optimize the whole vehicle model by using the noise optimization scheme. Therefore, the technical problems of high time cost and high labor cost due to the fact that the real vehicle needs to be tested for multiple times to determine the optimization scheme according to the test data in the related technology are solved.

A method, apparatus, non-transitory computer readable medium, and system for generating synthetic image includes obtaining an input prompt describing a first element and a second element. In some cases, an image generation model generates an intermediate output based on the input prompt and optimizes the intermediate output based on an attention contrast loss to obtain an optimized intermediate output. For example, the optimized intermediate output represents the first element at a first location and the second element at a second location. The image generation model generates a synthetic image based on the optimized intermediate output. The synthetic image depicts the first element at the first location and the second element at the second location.

The application relates to the technical field of air conditioners, and discloses a silencer. The silencer provided by the application comprises a silencer body, a first pipe, a second pipe and an inner connecting pipe. The first pipe comprises a first folding pipe section; the second pipe is fixed at two ends of the silencer body with the first pipe; one end of the inner connecting pipe is inserted into the first pipe and penetrates into a silencing chamber, and the other end of the inner connecting pipe is used for connecting to a refrigerant circulating system; a positioning point is arranged at the middle part of the inner connecting pipe, and the positioning point is fixedly connected with the end part of the first pipe; the first folding pipe section can be expanded or contracted, and the first folding pipe section is used for adjusting the depth of the inner connecting pipe penetrating into the silencing chamber. The length of the inner connecting pipe in the silencing cavity is variable, the length of the silencing chamber and the length of the inserted inner connecting pipe are used for judging the frequency and the sound elimination amount which need to be eliminated by the system, so that the problem that the silencer needs to be frequently replaced in the process of optimizing and improving the air conditioner noise is solved. The application further discloses an air conditioner.

The invention relates to a double-engine propeller noise prediction and optimization method based on an analytical method, and belongs to the field of aviation acoustics. The method disclosed by the invention comprises a noise prediction method and a noise optimization method, the noise prediction method establishes a noise prediction model based on an improved propeller characteristic theory, and the noise optimization method comprises the following steps: S1, establishing a noise objective function based on the noise prediction model, solving a first-order partial derivative and a Hessian matrix of the noise objective function, and obtaining a first-order partial derivative of the noise objective function; the method comprises the steps of S1, providing a noise target function, S2, providing a new Euler formula transformation method, and converting a triangular polynomial corresponding to a first-order partial derivative of the noise target function into a zero-equivalent algebraic polynomial, S3, analyzing and solving the zero point of the algebraic polynomial by using a Fourier-Frobenius matrix method, and S4, realizing synchronous maximum noise reduction control. According to the method, an improved propeller characteristic value theory and a Hessian + Fourier-Frobenius matrix method are used for rapidly and accurately solving an analytical solution of the minimum noise, and the method is combined with synchronous control, so that the limitation on the time and precision of predicting the minimum noise is overcome.

The invention relates to the technical field of data processing, in particular to a data processing method of an intelligent security comprehensive management and control platform. The method comprises the following steps: acquiring environment smoke concentration data; the method comprises the following steps: by taking data acquired in real time as real-time data, forming an observation interval by # imgabs0 # data before acquiring the real-time data; determining an oscillation factor of the real-time data according to the fluctuation condition of the data in the observation interval; calculating a noise optimization factor of the real-time data; taking a normalization result of a product of the noise optimization factor and the oscillation factor as a scale adjustment factor; adjusting a preset initial scale difference through the scale adjustment factor to obtain an optimal scale difference; and dividing a plurality of scales based on the optimal scale difference to calculate a multi-scale entropy to monitor the environment. The method has the effect of improving the system calculation efficiency and the monitoring accuracy.

The invention provides a data set knowledge density evaluation method and device based on noise optimization type data set distillation, and relates to the technical field of data processing.The method comprises the steps that on the basis that data set distillation is carried out based on a diffusion model, DDPM denoising is carried out batch by batch, performing random noise optimization in each step of denoising of each batch based on distribution alignment loss, maximum occupancy loss and authenticity constraint loss, and iteratively obtaining a final synthetic sample set of the last step of denoising of each batch; obtaining a distillation data set according to the final synthesis sample set of the last step of all batches of de-noised samples; the test performance of the classifier trained by the distillation data set is the same as that of the classifier trained on the original data set; and determining the knowledge density of the original data set according to the total data number of the distillation data set and the total data number of the original data set. According to the invention, the accuracy of knowledge density evaluation of the data set is improved.

The invention discloses an initial noise optimization method of a new view angle synthesis model based on a diffusion model, belongs to the diffusion model, and can be used in the field of new view angle synthesis. According to the optimization method, firstly, a machine learning network is provided, and the network is composed of an encoder and a decoder; then training an encoder-decoder network according to initial noise-optimized noise pair acquired by a new view angle synthesis model based on a diffusion model; and finally, inserting the trained encoder-decoder network into the new visual angle synthesis model. The invention aims to change initial noise into optimized noise through network action, so that a model generation result is improved on the premise of not finely tuning a model.