120 results about "Deviation vector" patented technology

The invention relates to the technical field of production process control, in particular to an MES system management and control method and system based on precision mold production and manufacturing, and the method comprises the following steps: obtaining a measurement point cloud and a theoretical model, calculating rigid body transformation matrix registration to generate a registration point cloud, searching a nearest projection point, measuring geometric deviation, and constructing a three-dimensional deviation vector; and calculating the actual residual material thickness, screening and generating margin insufficient risk points, analyzing a numerical control code to extract a coordinate instruction, retrieving a nearest risk point, calculating a compensation distance, and adjusting coordinates to generate a self-adaptive finish machining program. According to the method, the three-dimensional deviation vector is constructed to represent the entity state, the risk area is screened in combination with the machining allowance, the compensation distance is calculated according to the actual residual thickness, and the instruction is dynamically adjusted to generate the self-adaptive program, so that the path is automatically corrected according to the shape of the workpiece, and the overcut risk caused by insufficient allowance is effectively avoided; and accurate matching between the machining process and the workpiece state is ensured.

The invention relates to the technical field of vehicle control, and discloses a vehicle positioning and control method and system based on machine vision, and the method comprises the steps: building a vehicle kinematics prediction model, and generating a reference pose sequence according to a driving instruction through the vehicle kinematics prediction model; recording a first reference pose during image acquisition; calculating a deviation vector between a positioning result and the first reference pose when positioning calculation is completed; determining a dynamic damping coefficient according to a direction included angle between the velocity vector and the deviation vector, generating a compensation gain sequence, and converting the compensation gain sequence into a compensation torque of a current loop given value; according to the method, the effective decoupling of the sensing time delay of the sensor and the dynamic response of the carrier is realized, the phase lag is reduced on the premise of ensuring the momentum continuity of the controlled object, the instantaneous control oscillation of positioning correction is eliminated, and the mechanical stress impact of an execution mechanism is inhibited.

The invention relates to the technical field of motion control, in particular to a motion trail real-time optimization analysis method and system based on multi-modal perception, and the method comprises the following steps: a multi-sensor array obtains position and speed visual features, carries out the coordinate transformation, extracts a direction amplitude, and builds a multi-modal perception data set; the method comprises the following steps: inputting Kalman filtering prediction trajectory comparison deviation to generate a trajectory deviation vector field, calculating a deviation gradient, screening an over-threshold point marking time sequence, generating a dynamic weight correction parameter through weighted least square distribution normalization, and outputting a real-time optimization trajectory scheme through weighted correction fusion coordinates. The position information and the velocity vector of a moving target are obtained in real time by fusing multi-modal sensing data, a state estimation model is combined to predict a trajectory, a trajectory deviation vector field is generated, and a deviation point is subjected to weighted correction, so that the trajectory optimization precision is effectively improved, the weight is dynamically allocated to optimize the priority, and the trajectory optimization precision and the system stability are improved. And the self-adaption and robustness in a complex environment are enhanced.

The invention relates to the technical field of numerical control machining and automatic control, in particular to a numerical control machining self-adaptive control system and method based on multi-source data, and the method comprises the steps that FPGA hardware synchronously collects main shaft current, servo current of each feed shaft, cutting vibration and grating ruler pulse, and time bases are aligned and packaged into a machining multi-dimensional state matrix; the multi-physical field error is decoupled through a three-layer cascade machine tool dynamics model, a tool nose contour deviation vector is solved, and servo lag and cutting force deformation factors are fused to construct a dynamic constraint boundary. When the deviation amplitude exceeds the limit, generating a speed feed-forward gain compensation instruction based on an inverse control model to perform instant suppression; when it is monitored that oscillation is eliminated and returns to a steady state, full-time-domain data stream interception logic is triggered, and a process optimization instruction is generated through backtracking. According to the method, the problems of nonlinear time-varying error decoupling and real-time compensation lag under multi-physics field coupling are solved, and the dynamic precision and process stability of numerical control machining under complex working conditions are remarkably improved.

The invention relates to a loose-leaf notebook inner page hole site processing self-adaptive positioning method and system and electronic equipment, and the method comprises the steps: obtaining the morphological feature information of the edge of a to-be-processed paper pile, and determining a datum reference line and an actual edge contour curve according to the morphological feature information; and based on a deviation mapping model, determining a pose deviation vector according to the datum reference line and the actual edge contour curve, generating a positioning adjustment instruction according to the pose deviation vector by utilizing a self-adaptive compensation algorithm, and driving a positioning mechanism to carry out pose correction on the to-be-processed paper pile. And aligning the hole site machining center with the actual edge profile curve. Accurate sensing and self-adaptive compensation of the shape of the edge of the inner page of the loose-leaf notebook are achieved, the problem that traditional mechanical positioning is poor in adaptability is effectively solved, high precision and high quality of hole site machining are guaranteed, and the yield is increased.

The invention discloses an intelligent customer service question and answer optimization method and system based on large model driving, and relates to the technical field of data processing.The method comprises the steps that a semantic flow self-supervision state machine is used for generating a semantic attribution graph and extracting core semantic factors, reverse semantic generation and cross-factor compensation are executed, and a structured target semantic set is constructed; a multi-branch task state diagram is constructed through behavior path inversion, and after a main task state is recognized, two-dimensional semantic projection is carried out, and a strategy tensor is generated; generating a polymorphic answer and constructing a reply text based on the strategy tensor, the main task state and the semantic set; and calculating a deviation vector through a user feedback path and updating the large model. The technical problem that an existing intelligent customer service system is low in semantic understanding precision, single in answer and lack of individuation, and consequently the question and answer response speed and precision are low is solved, and the technical effect that the semantic understanding precision and the answer diversity are improved through a dynamic optimization mechanism driven by a large model, so that the question and answer response speed and precision are improved is achieved.

The invention discloses a tunnel mining engineering blasting effect evaluation and analysis method based on the Internet, and relates to the technical field of tunnel mining engineering, and the method comprises the steps: carrying out the synchronous fusion of a multi-modal feature set, and generating a multi-dimensional blasting feature representation; performing multi-scale time sequence decomposition and spectrum energy clustering on the multi-dimensional blasting characteristic representation to generate a blasting response mode, and performing statistics and spectrum analysis to generate a blasting effect evaluation index set; comparing the blasting effect evaluation index set with a standard template in a historical blasting engineering database to generate a deviation reference, and calculating a current blasting effect deviation degree by adopting a Bayesian dynamic inference method to generate a deviation vector; performing dynamic weight adjustment and feature contribution correction on the deviation vector to generate a comprehensive blasting effect scoring result; and analyzing and summarizing the comprehensive blasting effect scoring results to generate a blasting effect evaluation report. The reliability of blasting quality control and safety early warning is improved.

The invention discloses a tunnel engineering blasting effect evaluation and analysis method based on big data, and relates to the technical field of geotechnical engineering monitoring, and the method comprises the steps: constructing a blasting digital twinborn body based on a blasting energy thermodynamic diagram, reconstructing a rock mass fracture extension three-dimensional characteristic field according to space energy density distribution, and generating a triangular mesh model for simulating a blasting contour; aligning the real spot cloud data of the tunnel blasting section with the triangular mesh model of the blasting contour to obtain a contour deviation value and a three-dimensional contour deviation field, identifying a contour deviation sagittal plane domain and marking a stake mark interval position; and based on the space energy density distribution and the contour deviation value, predicting a blasting effect comprehensive score, and generating a blasting effect analysis report. According to the method, multi-dimensional intelligent evaluation of the blasting effect is realized through learnable weight parameters and a Sigmoid gating mechanism, the overall effect can be reflected, the cause of a specific problem can be positioned, and a continuously optimized evaluation system is formed.

The invention discloses a personality adaptive training and dynamic evolution method of a role large model and a related product. The method comprises the steps of obtaining interaction log data when a self-adaptive training triggering condition is met; scoring the interaction log data to obtain an index score vector, comparing the index score vector with a target score vector to obtain a deviation vector, and summarizing the deviation vector to form a deviation report; constructing a training sample based on the deviation report; on the premise of freezing trunk parameters of the basic language generation model, executing small-step increment updating on an evolvable personality vector and / or a scene sub personality increment module set in the personality configuration file, and applying core personality attribute anchoring constraint, distribution difference constraint and rule constraint to obtain a target personality configuration file; and when the target personality configuration file passes the security verification, writing back and storing in a versioning manner, otherwise, rolling back to update the current personality configuration file. According to the method, personality controllable evolution is realized in multi-scene continuous interaction, and the risk of personality drift and cross-scene inconsistency is reduced.

The invention relates to the technical field of control and adjustment, and discloses a real-time control system and method for the production progress of an intelligent manufacturing platform, and the system comprises an information extraction module, a data packaging module, a state sequence module, a progress association module, a progress deviation module and a cooperative control module, and is used for obtaining multi-source heterogeneous data and a scheduling resource set of the intelligent manufacturing platform; packaging the multi-source heterogeneous data to obtain a production live set; mapping the production live set to a production element sequence of an intelligent manufacturing platform to obtain a production state sequence; extracting an evolution vector and a scheduling vector in the production state sequence to obtain a progress association vector pair; forward iteration is carried out on the progress offset between the progress association vector pairs based on process parameters in the scheduling vector, and a deviation vector of the intelligent manufacturing platform is obtained; performing collaborative optimization on the scheduling resource set based on the deviation vector to obtain a collaborative control sequence of the intelligent manufacturing platform; according to the invention, the accuracy and efficiency of real-time control of the production progress of the intelligent manufacturing platform can be improved.

The invention relates to the technical field of robot control, and discloses a power station narrow space inspection robot control method. According to the method, when a preset inspection triggering condition is met, a historical operation state of a target inspection robot at a specified historical moment is obtained, the historical operation state comprises historical accumulated operation duration and historical environment parameters, and meanwhile, a real-time operation state is obtained, and the real-time operation state comprises real-time accumulated operation duration and real-time environment parameters. Through analysis of historical and real-time operation states, an operation state deviation vector is constructed, and a corresponding operation deviation measurement set is retrieved. When the set distribution dispersion is larger than or equal to a dispersion threshold value, the ratio of the historical duration from the storage starting moment to the specified historical moment to the current duration from the storage starting moment to the current moment is calculated and serves as a control parameter adjusting proportion, and then control parameters of the target inspection robot are updated according to the proportion.

The invention discloses an attitude control method and system for continuously locking a ground target by an unmanned aerial vehicle, and the method comprises the steps: obtaining a target position and an unmanned aerial vehicle attitude data stream, carrying out the coordinate transformation to generate a target attitude relation identifier, and carrying out the deviation vector decomposition to form a visual axis deviation feature set; disassembling the image into a fast and slow response chain through attitude mapping processing, repairing a fracture node, constructing an attitude correction track, extracting a control feature mode, generating a response feature factor, and constructing an attitude compensation constraint matrix; frequency domain decomposition and cross coupling calculation are carried out on the optical axis deviation feature set to determine the coupling correlation degree, and a superposition disturbance sequence is identified and dynamically reconstructed to generate a compensation gain sequence; executing stability evaluation to determine time sequence distribution, identifying abnormal disturbance characteristics and extracting key moment control characteristics; the constraint matrix and the gain sequence are fused to generate a fusion control law, a dynamic priority control coefficient is determined, a target continuous locking attitude control instruction is output, and continuous and stable locking of the moving target is achieved.

The invention discloses a wire harness processing control method and system based on artificial intelligence, and the method comprises the steps: collecting sensor array data, carrying out the normalization processing, forming a standardized parameter set, precisely recognizing an abnormal mode through a classification algorithm, marking a potential fault point, extracting a time sequence feature sequence, and generating a prediction deviation vector through a prediction algorithm, analyzing dominant influence factors, grouping similar deviation modes through a clustering algorithm, and judging process adjustment requirements; and an optimization instruction sequence is generated, a production scene is matched based on historical data, the adjusted parameter configuration is dynamically injected into a control system, and the parameter model is continuously optimized through real-time monitoring and feedback circulation. According to the method, through data-driven anomaly detection and prediction adjustment, the stability and efficiency of the manufacturing process are remarkably improved, the fault rate is reduced, and intelligent machining state optimization is achieved.

The invention relates to the technical field of industrial process control, and discloses a plastic product large-scale production multi-parameter cooperative control system and method. A deviation calculation module; an adaptive compensation module; and a set value updating module. The method comprises the following steps: injecting a probe signal in a preset non-sensitive stage, synchronously acquiring a response signal, and calculating a process response feature vector; calculating the deviation between the measurement vector and a preset reference vector to generate a measurement deviation vector; determining a target compensation vector through deviation attribution in combination with prediction drift, updating a process response Jacobian matrix online, and calculating a multi-parameter cooperative adjustment amount; and updating the machine set value of the next period according to the adjustment amount. According to the method, the problems of control lag and parameter coupling are solved through prospective compensation and self-adaptive updating, and the control precision and robustness of multiple parameters in large-scale production of plastic products are improved.

The invention provides an arc-shaped lithium battery intelligent forming method and system fusing digital twinning, and relates to the technical field of digital twinning. According to the method, a twin covering geometry, material evolution, heat engine coupling and equipment dynamics is established, unified modeling is carried out on a forming target, process data and image data, millisecond-level collection is carried out on the edge side, deviation vectors are output, parameter identification and time delay compensation are driven, a formula is generated through online optimization, and closed-loop execution is carried out; and when the judgment result is unqualified, triggering re-calibration and re-optimization, and filing and updating design parameters, formulas and measurement and judgment results to form a cross-batch learning and knowledge base. According to the invention, experience dependence and tracing fracture are overcome, and real-time controllability, full-link traceability and cross-batch optimization are realized.

The application relates to a dynamic vision-based puncture deviation correction method, device, equipment and storage medium, which comprises collecting image data and spatial position data of a to-be-punctured object, then extracting target positioning information from the image data and constructing a corresponding three-dimensional deformation model; determining corresponding offset data according to the three-dimensional deformation model, fusing the target positioning information and the offset data to generate a deviation vector; based on the deviation vector, performing fuzzy control rule calculation and combining historical control data to generate a control output; by analyzing the change trend of the deviation vector at multiple time points, predicting the offset direction and amplitude of a preset time period to generate a feedforward compensation amount, and then generating a compensation control instruction; generating three-order trajectory information of a puncture path according to the compensation control instruction to complete dynamic deviation correction control according to the three-order trajectory information. The application has the effect of improving the precision of an automatic puncture system.

The application provides a kind of castings logistics package automatic marking system, it is related to automatic detection technical field, including data acquisition module obtains marking deviation detection parameter and verifies validity;Deviation characteristic calculation module calculates the instantaneous deviation distance of actual marking point, and calculates dynamic deviation ratio in combination with package size;Window construction module determines whether marking position is qualified according to dynamic deviation ratio, constructs deviation sliding window and inputs the instantaneous deviation distance and coordinate deviation vector of qualified event;Trend evaluation and early warning response module calculates system drift coefficient based on the data in sliding window, determines whether marking deviation exists systematic cumulative trend and generates early warning signal.The application eliminates the influence of package specification on deviation evaluation by dynamic deviation ratio, realizes from single-point qualification determination to long-term trend monitoring by sliding window and drift coefficient, identifies the cumulative drift of marking equipment, avoids batch marking deviation accident, and improves the quality control ability of marking process.

The invention discloses a generative model optimization system and method applied to closed-loop verification feedback in the field of artificial intelligence, and the method comprises the steps: outputting an initial generation result through a generative model, inputting the initial generation result into an in-vitro verification unit, and carrying out the functional testing, thereby obtaining a multi-dimensional performance evaluation signal; calculating a dimension-by-dimension deviation vector between the target threshold value and the target threshold value, and taking the dimension-by-dimension deviation vector as structured feedback to drive model parameters to be directionally updated; according to the method, a real-time closed-loop feedback path of physical / simulation environment and model training is constructed, so that the generation process directly takes real performance response as a supervision signal, the generation precision, the target adaptation degree and the multi-index collaborative optimization efficiency are remarkably improved, and the method has the advantages that the real performance response is directly taken as a supervision signal. And meanwhile, the verification cost is reduced through geometric pre-check and a track caching mechanism, and a traceable and extensible intelligent generation optimization normal form is formed.

This invention discloses a product data verification method and system based on big data analysis. The method includes: performing initial verification to obtain initial verification data; performing anomaly detection and determining the verification time period and verification evaluation data based on the anomaly detection results; extracting data centers for products in the same category; calculating the centrality of each verification evaluation data point to obtain a data center degree vector; calculating the confidence of each verification evaluation data point based on a time-series dynamic confidence decay model to obtain a confidence matrix; extracting and normalizing user feedback data to obtain a user feedback vector; calculating the correlation between each verification evaluation data point and the remaining data and calculating the correlation deviation vector; constructing a semi-supervised adversarial verification prediction model; and inputting the product data to be verified into the semi-supervised adversarial verification prediction model to obtain the verification results. This method not only improves the efficiency and accuracy of product data verification but also has good interpretability and can be directly applied to product data verification systems.

This invention relates to the technical field of paper bag production, specifically to a machine learning-based method for paper bag gluing positioning and adhesion control, comprising the following steps: S1, generating original grayscale data; S2, generating the maximum structural response amplitude at multiple frequency levels and the local phase information of each pixel; S3, generating a high signal-to-noise ratio structural amplitude map; S4, generating a structural feature map; S5, generating glue path segmentation results; S6, generating control commands based on the deviation vector; by calculating the multi-frequency structural synchronization index through frequency domain phase analysis, systematic positioning deviations caused by anisotropic interference are eliminated; a texture-independent structural feature map is generated, so that the geometric segmentation network does not need to be retrained when changing paper batches, completely solving the domain drift problem; combined with fast adaptive direction decomposition and lightweight network design, the real-time closed-loop control requirements of high-speed production lines are met, significantly improving gluing accuracy and production line adaptability.

This invention discloses a manufacturing process optimization method based on defect pattern mining, relating to the field of data processing. The method includes: obtaining defect indicator detection data; performing a first-level de-randomization analysis based on the detected defect indicator type data to obtain high-frequency detected defect indicator types; performing a second-level de-randomization analysis on the defect indicator deviation vector data to obtain high-frequency defect indicator deviation vectors; obtaining production control attributes from the first node to the Nth node; performing correlation analysis to obtain correlation control attributes; matching control variables on the high-frequency defect indicator deviation vectors to obtain process calibration attributes; and sending the results to a quality control terminal. This method solves the technical problem of existing production process optimization methods lacking in-depth analysis of quality-related causes, thus failing to fundamentally optimize the production process. It achieves the technical effect of precise control and optimization of the production process by uncovering the root causes of product quality problems.

The application discloses a wafer high aspect ratio Bosch etching parameter self-adaptive regulation method, which comprises the following steps: step one, collecting wafer to-be-processed structure data, Bosch etching process data and stage profile data, and preprocessing to generate standardized data; step two, performing period phase segmentation to generate period state slices; step three, performing period memory constraint state decoupling to construct a state variable set and a control input vector; step four, performing mechanism coupling candidate expansion to generate a nonlinear candidate function library; step five, calculating state derivatives and identifying through SINDYc to generate a sparse dynamic control equation set; step six, rolling predicting profile evolution and generating a profile deviation vector; and step seven, correcting parameters according to the profile deviation and iteratively updating to output final process parameters and etching profile results. The application realizes precise regulation of high aspect ratio Bosch etching parameters, and improves profile stability and processing consistency.

Embodiments of the present application disclose a virtual object control method and device, electronic equipment and a storage medium; the method comprises: acquiring a direction of a sight line and a starting point of the sight line, and displacement and speed of a virtual object; calculating a deviation vector between the virtual object and the starting point of the sight line according to the starting point of the sight line and the displacement of the virtual object; decomposing the speed into speed components in a plurality of specified directions according to the deviation vector; calculating acceleration in the specified directions; fusing the acceleration in all the specified directions to obtain a target acceleration of the virtual object; and controlling the virtual object to move along the direction of the sight line based on the target acceleration. Based on simple speed decomposition and acceleration calculation, the control process of missile guidance is simplified, and the calculation amount in control is reduced, so that the missile guidance can be efficiently applied in games to realize control of the virtual object.

This invention discloses a smart park load forecasting method integrating seasonal and solar term features, comprising: generating a feature input vector and a meteorological-solar term deviation vector based on current input data; a trained weight calculation module outputting a weight vector and a bidirectional dynamic hidden state vector based on these two vectors; a trained gating network outputting a fuzzy probability distribution based on the weight vector and the bidirectional dynamic hidden state vector; and a trained fuzzy boundary hybrid expert network outputting park load forecast data for a preset future time period based on the weight vector, the bidirectional dynamic hidden state vector, and the fuzzy probability distribution. The expert network includes multiple trained seasonal and solar term sub-networks, each of which uses the knowledge learned during training to predict park load data. This invention can deeply capture the nonlinear spatiotemporal relationships of multi-source data, significantly improving the prediction accuracy, convergence speed, and robustness of the model under complex and variable climate conditions.

The application discloses a product reconstruction method and system based on three-dimensional scanning and CAD surface deviation distribution, and relates to the technical field of three-dimensional digital measurement. The method firstly carries out CAD pretreatment on each part to obtain a reference CAD model and a finite element model, and scans and acquires single-piece scanning point clouds. The point clouds are accurately aligned with the reference CAD model through single-piece registration, and then the projection point corresponding to each scanning point and the deviation vector thereof are calculated. After physical assembly, the part point clouds are positioned to a product coordinate system through product registration. Then, the surface nodes of the finite element model are classified based on the deviation vector. Finally, the corrected finite element model and the reference CAD model are repositioned to the product coordinate system, and a digital prototype and a high-fidelity finite element analysis model integrating actual geometric deviation and assembly information are output. The application significantly improves the convenience, precision of digital measurement and fidelity of finite element simulation.

The application provides a physical mechanism coupling data-driven fast prediction method, and belongs to the technical field of disaster prediction.The application extracts an invariant topological feature vector through topological data analysis driven phase space reconstruction and inputs a physical constraint identification model to obtain a conservation law deviation vector and a physical feasible region boundary parameter, adopts a multi-resolution adaptive grid technology to perform multi-scale decomposition, extracts long-term evolution trend features and spatial local features through a full connection layer network and a convolution layer network, combines the physical feasible region boundary parameter to perform projection gradient descent iterative optimization to generate a preliminary prediction field, performs Bayesian uncertainty quantization, adjusts an artificial intelligence model regularization coefficient according to a disaster precursor identification function value, and outputs a graded early warning result, so that the technical problem of insufficient prediction accuracy caused by insufficient coupling of a physical constraint and a data-driven model in a disaster prediction system is solved.

This invention relates to the field of process control technology, and discloses an automatic control method and system for actual drilling trajectory comparison deviation. The method includes: performing coordinate analysis on the inclination angle and azimuth angle at the well depth and the spatial coordinate sequence of the designed trajectory at the well depth to obtain a spatial coordinate sequence; performing sliding window comparison analysis on the spatial coordinate sequence and the spatial coordinate sequence of the designed trajectory at the same well depth to obtain a multi-scale deviation feature set; performing nonlinear mapping inference on the multi-scale deviation feature set to obtain a dynamic control target vector; performing correlation analysis on the tool face angle in the dynamic control target vector and the measured tool face angle to obtain a bottom-level drive command sequence; performing spatial registration evaluation on the spatial coordinate sequence and the preset design trajectory to obtain a spatial deviation vector sequence; and adaptively adjusting the bottom-level drive command sequence to obtain a target drilling scheme. This invention can improve the efficiency of automatic control of actual drilling trajectory comparison deviation.

This invention discloses a smart simulation method and system for urban flood control. The method includes: acquiring water level monitoring data from the urban flood control monitoring network and constructing a real-time deviation vector based on the model status; calculating the observation information value index characterizing the contribution of each monitoring point to parameter correction based on the uncertainty of model parameters and their sensitivity to monitoring points; performing online rolling correction of hydrodynamic model parameters using weighted or filtered indicators; calculating a dynamic risk index reflecting the degree of disaster threat based on the simulation results, and determining the grid spatial resolution of each region within the computational domain based on the observation information value index; and performing simulation on a grid with the corrected parameters at that resolution to output flood evolution results. This invention achieves spatiotemporal adaptive optimization of computational resources and correction strategies by actively sensing the value of observation information and dynamic risk distribution, solving the technical problems of low data utilization efficiency and loss of focus in simulation in traditional methods.

The invention relates to the technical field of data processing, in particular to a peptide milk powder production line monitoring method and system based on multi-source data fusion. The method comprises the following steps: acquiring real-time working condition data and off-line detection data of a peptide milk powder production line; calculating a relative error of the offline data and the extended Kalman filtering output, and mapping the relative error into a state drift divergence; calculating a blind area heat cumulative gradient based on the raw milk density, the real-time temperature and the pipeline flow velocity; constructing a time confidence coefficient smooth surface in combination with fluid residence duration to generate a backtracking migration gain coefficient; and extracting a historical deviation vector, and calculating a dynamic compensation item by using a gain coefficient to correct a filtering posteriori state updating equation so as to output a corrected real-time state of the peptide milk powder production line. The method can more accurately evaluate and predict the deviation, scientifically analyze the heat exchange accumulative transfer effect, ensure time-space reasonable attenuation of the historical deviation, improve the monitoring accuracy and real-time performance, effectively balance the sterilization effect and peptide activity protection, and reduce the production loss.