824 results about "Multi variable" patented technology

A method for multivariate time-series long-term forecasting based on multi-scale temporal feature enhancements, includes a time-series forcasting model TFEformer. The model utilizes a multi-branch structure and a patch-series attention mechanism to extract global and local time-series features at multiple temporal scales, and designs an adaptive feature fusion mechanism to achieve adaptive fusion of multi-scale temporal features. It employs an variate-wise attention mechanism and a redesigned gated feedforward network to perform feature fusion among multivariate variables and within the time-series, respectively. The time-series forcasting model TFEformer proposed by the present invention significantly improves the prediction of long-term trends in time-series and enhances the fitting ability for short-term local fluctuations, comprehensively increasing prediction accuracy across different prediction time lengths in multivariate time-series forcasting tasks.

The invention discloses an interpretability prediction analysis system for multivariable time series data. According to the method, the prediction precision and the decision support capability of the complex time series data are remarkably improved through multi-module cooperation. Firstly, an adaptive learning optimization module dynamically adjusts model parameters and a prediction strategy, so that the model can quickly adapt to time-varying characteristics of data distribution, for example, when a causal relationship between variables suddenly changes, the weight of latest data is automatically enhanced, and historical noise interference is reduced. The dynamic causal interpretation engine tracks the influence intensity and hysteresis effect of key variables in real time, converts traditional black box prediction into a traceable causal relationship chain, and helps a user to intuitively understand driving factors of a prediction result, for example, it is identified that prediction value sudden increase in a certain period is mainly derived from hysteresis effect accumulation of an upstream variable A.

The invention belongs to the technical field of intelligent traffic, and particularly relates to a Mama-based traffic flow prediction method, which comprises the following steps of: collecting and preprocessing a data set; static graph construction and spatial feature extraction; performing space-time dual-token decoupling embedding; carrying out dual-branch Mama modeling; performing space-time-variable feature fusion; and carrying out multi-step prediction, outputting a prediction result, and finally, outputting a prediction value of a multi-step sequence through projection mapping. The static graph convolutional network extracts spatial correlation from the inherent topology of the road network, and the double-branch Mama architecture captures the time dependence and variable correlation of the traffic flow. According to the space-time double-token mechanism, the independent influence and the synergistic effect between variables are explicitly distinguished by decoupling the action boundaries of multi-source factors such as traffic flow, environment variables and event interference, and the modeling robustness and the interpretability in a multivariable coupling scene are remarkably improved.

The invention discloses a multivariable source load data prediction method and system based on a pre-trained large language model, and the method comprises the steps: carrying out the textualized coding of a historical source load time sequence, forming a pre-trained large language model based on a multi-expert mixed fine tuning module through the mixed use of a low-rank matrix adaptation method and an adapter method, and obtaining a preliminary prediction time sequence; finally, filtering calibration is carried out, and a variable with higher correlation is selected as an advanced variable based on the cross correlation between the variable to be predicted and other variables; filtering is carried out on the advance variable data and the variable data to be predicted in the preliminary prediction time sequence, and then integration is carried out to obtain a calibration deviation of the preliminary prediction time sequence for calibration. Fine adjustment is carried out on the basis of a large language model, different types of source load sequential sequence prediction tasks can be completed under the conditions of small samples and multiple variables, prompting and guiding operation is carried out on the to-be-predicted variables through the lead-lag relation between the variables, and the source load prediction accuracy and reliability are improved.

The invention discloses a multivariable time series prediction method based on an NMF multi-scale lightweight space-time convolutional neural network, and belongs to the technical field of machine learning and deep learning application. The method specifically comprises the following steps: (1) collecting and processing multivariable time sequence data, and unifying the scale of the data; (2) decomposing the time sequence data by using NMF, extracting basis matrixes and coefficient matrixes of different scales, and constructing a multi-scale feature pyramid; (3) applying a space-time convolution layer on the basis of the output of the feature pyramid, introducing a Bayesian optimization algorithm, and adopting a feature fusion layer fused with a multi-scale attention mechanism; (4) dividing a training data set and a test data set, and carrying out model training; and (5) predicting future time sequence data by using the trained model and outputting an expected value. According to the method, a more powerful and more efficient multivariable time sequence prediction model is effectively constructed, and the method is suitable for analysis and prediction of multivariable time sequence data.

The invention discloses a package management and control method fusing multivariable dynamic analysis, relates to the technical field of package dynamic management and control, and covers a complete process from data acquisition to protection strategy implementation. Firstly, multi-source sensors are arranged at packaging and transportation nodes, stress, acceleration and environmental parameters are collected in real time, and multi-dimensional original data covering the whole process is formed; then, carrying out time domain alignment and feature extraction on the data by utilizing denoising filtering, wavelet decomposition and environment coupling methods to obtain structured data of a unified coordinate system; and accurately identifying a packaging fatigue risk through a microcrack jump function and an accumulated impact measurement model, and generating a crack propagation trend and failure prediction. And finally, according to the risk change of the section, executing a sectional protection strategy, and dynamically adjusting and optimizing a protection scheme to ensure efficient management and control. The limitation of traditional packaging control is overcome through nonlinear modeling and real-time adjustment and optimization, and the method has a wide application prospect.

The invention relates to the technical field of control, and particularly discloses a multivariable energy efficiency optimization control system for a heating furnace, which is used for solving the problems of local overheating, non-uniform temperature and difficulty in accurate positioning and compensation of heat loss in the operation of the existing cracking heating furnace. Comprising a parameter detection module, a multivariable coupling modeling and simulation module, an optimization control module and an execution and feedback module. According to the method, dynamic digital twinning is constructed through multi-modal online sensing and data assimilation, a Pareto frontier solution is generated based on model prediction control and improved NSGA-II parallel optimization, and the weight is adaptively adjusted; and when the hot spot / cold spot is triggered, a quadric surface fitting compensation strategy is implemented and issued for execution, so that high-precision simulation prediction, precise closed-loop control and real-time online energy efficiency optimization are realized.

An SCR flue gas denitration intelligent control method based on multivariable collaborative optimization specifically comprises the following steps: S1, collecting multi-key variable data of SCR flue gas denitration, processing the multi-key variable data, and storing the processed multi-key variable data into a historical database; s2, on the basis of the processed data, constructing a multivariable correlation model, and predicting operation states of the denitration system under different working conditions; s3, according to the actual operation condition of the system and the equipment performance, determining constraint conditions and optimization targets of SCR flue gas denitration; s4, performing multivariable collaborative optimization calculation in combination with the multivariable correlation model, the optimization target and the constraint condition, and searching an optimal control variable combination; and S5, intelligent control is implemented according to the optimal variable combination, and re-optimization is performed by monitoring feedback deviation in real time. The system can keep stable control precision in different operation scenes, and denitration efficiency fluctuation or parameter adjustment lag caused by sudden change of working conditions is avoided.

The invention discloses a multivariable time sequence prediction method based on GRU and a computer program product. The method comprises the following steps: firstly, introducing the dynamic characteristics of self-adaptive different time sequences of self-defined time sequence decomposition, and splitting an original sequence into trend, seasonal and residual components so as to reduce the data complexity and improve the interpretability; afterwards, a value embedding module is used for unifying feature representation so as to ensure that the model fully captures the time dependency relationship; in the modeling stage, the model adopts a multi-channel recurrent neural network to independently model the three types of components so as to reduce the interference between modes and improve the learning ability. In the prediction stage, a feature splicing strategy is adopted, information of each component is integrated, and richer time sequence representation is provided. In addition, a segmented prediction strategy is designed for the model, the prediction process is divided into multiple time periods, prediction information is combined, error accumulation is reduced, and the stability and robustness of long-sequence prediction are improved. Experimental results show that the prediction precision can be improved on different data sets.

The invention discloses a mapping prediction method and system based on a hardware resource load and system operation relationship. The method comprises the following steps: collecting hardware index data in real time when an operating system operates; the method comprises the following steps: preprocessing hardware index data, performing association analysis on each index data by adopting an association rule mining algorithm, generating association rules, evaluating the association rules, screening out the association rules meeting conditions, and constructing a multivariable association model; based on a correlation analysis result, a load prediction model is constructed and trained, and the trained model is deployed in the system for load prediction; and dynamically adjusting a system hardware resource allocation strategy according to a prediction result, carrying out task migration and load balancing according to a predicted load condition, and automatically adjusting hardware resources through an automatic script or a scheduling tool. Through real-time monitoring and deep correlation analysis of indexes such as the CPU, the memory, the storage I / O and the network bandwidth, the resource utilization rate and the operation efficiency of the system are improved, and the stability and the reliability of the system are improved.

The invention relates to the technical field of multivariable analysis, and discloses a method and a device for improving the strength of flow-state solidified soil based on multivariable analysis, and the method comprises the following steps: collecting a material proportion parameter matrix, a stirring parameter matrix and an environment monitoring data matrix in flow-state solidified soil equipment, and executing feature extraction and feature integration; obtaining a target feature vector; performing intensity prediction of the time sequence on the target feature vector to obtain an intensity prediction value of the flow-state solidified soil; sensitivity analysis is carried out based on the intensity predicted value, and an intensity improvement potential index, a parameter adjustment stability index and an energy consumption influence index are obtained; according to the strength improvement potential index, the parameter adjustment stability index and the energy consumption influence index, adaptive parameter optimization is carried out through discount inverse reinforcement learning, and a target control parameter set is obtained. And the strength and the quality uniformity of the flow-state solidified soil are effectively improved while the operation stability is guaranteed.

The invention discloses a multivariable time series prediction method based on Patching and multi-scale feature extraction, and the method comprises the steps: dividing obtained multivariable time series data into different independent channels for a multivariable long-term time series prediction task, sharing the same converter backbone network, but enabling a forward process to be independent; the method comprises the following steps of: firstly, performing a Patching operation on a long sequence from a time domain angle, segmenting a single variable of each channel into a plurality of short fragments, and transmitting the short fragments into a Transform framework; in the aspect of a frequency domain, a multi-scale convolutional network is adopted to capture features of different frequency ranges, and meanwhile, a scale attention mechanism is introduced to adaptively weight and fuse features of different scales; and carrying out feature fusion on the features extracted from the frequency domain and the time domain so as to realize accurate modeling of periodic and long-term modes. According to the method, experimental analysis is carried out on eight disclosed data sets, and the result shows that compared with existing models, the model provided by the invention has higher accuracy on prediction of future values in a selected scene.

The invention relates to the technical field of ocean data interpolation filling, in particular to a space-time deficiency filling method and system based on context association and physical guidance. The method comprises the following steps: acquiring seawater dissolved oxygen data and context data; multivariable space-time dependence extraction is carried out based on the obtained seawater dissolved oxygen data and context data; gaussian noise diffusion is carried out based on the obtained seawater dissolved oxygen data; noise prediction is carried out based on double-view space-time correlation; and the prediction error is constrained based on the joint loss function. According to the method, a physical consistency constraint mechanism based on a partial differential equation is introduced in a model training process, so that model output better conforms to a physical coupling rule among variables in a marine environment. The constraint effectively inhibits non-physical fluctuation possibly occurring in the interpolation result, enhances the physical credibility and interpretability of the result, and provides a more reliable data basis for subsequent scientific analysis and process modeling.

The invention provides a multivariable multi-step runoff prediction method based on dynamic variational mode decomposition, and belongs to the technical field of drainage basin runoff prediction.The method comprises the steps that various hydrometeorological data of a research drainage basin are collected; decomposing an original runoff sequence by adopting a dynamic variational mode decomposition strategy VMD, performing sliding window slicing processing on drainage basin underlying surface and meteorological data and upstream hydrologic station runoff data, and combining target hydrologic station runoff data subsequences obtained by decomposition into a first characteristic variable; meteorological factor screening is carried out on the meteorological station data, a screening result is used as a second characteristic variable, and the first characteristic variable and the second characteristic variable are combined together to form an input characteristic variable; a sparrow search algorithm SSA is adopted to optimize parameters of a long short-term memory network LSTM to construct the long short-term memory network LSTM, and future runoff is predicted. According to the method, the defect that future forecast information is introduced by binding decomposition is effectively overcome, and the applicability of the decomposition-integration forecast model in actual engineering is enhanced.

The invention relates to the technical field of meteorological monitoring and climate prediction, in particular to a Qinghai-Tibet Plateau composite extreme climate event attribution evaluation method. The method comprises the steps that ground observation, remote sensing and reanalysis data are integrated through a multi-source data dynamic space-time weight fusion technology, and abnormal value correction and non-uniform interpolation are achieved; identifying a composite event by adopting a multivariable combined extreme index and a space-time coupling graph model and generating a structured label, wherein the structured label comprises strength, range, duration and evolution path; constructing a multi-scale causal network to analyze the contribution of the driving factor, and implementing physical constraint disturbance based on causal weight; recovering high-resolution response by using a Bayesian agent model and combining topographic constraint random downsampling, and deducing spatio-temporal evolution through an event propagation network; and a kernel polynomial hybrid uncertainty propagation framework is adopted to generate a probabilistic scene set, and multi-level risk early warning and dynamic knowledge base optimization are realized. According to the invention, the attribution precision and early warning efficiency of plateau composite extreme events are comprehensively improved.

The invention relates to a machine learning-based scouring damage bridge buckling load prediction method and system, and belongs to the technical field of machine learning and structural engineering. The prediction method comprises the steps of obtaining original data, preprocessing the original data, constructing a LightGBM machine learning model, and evaluating and verifying the constructed model by adopting multiple performance evaluation indexes. According to the method, key variables such as bridge structure parameters and foundation soil physical property indexes are comprehensively considered, a large sample database covering a large number of scouring combinations is constructed, and the critical buckling load of the structure is accurately calculated as an output target by using a pier-bearing platform-pile structure system buckling analysis method after scouring damage based on an energy method. A multivariable nonlinear mapping relation is established in combination with an ensemble learning algorithm, traditional complex calculation is replaced, and the method has good engineering adaptability and universality and can be suitable for rapid prediction of the buckling bearing capacity under different parameter bridges, different foundation forms and complex scouring conditions.

The invention belongs to the technical field of power system load prediction, and relates to a long-range multivariable load prediction method and system based on time-frequency domain collaboration, and the system carries out the normalization and stabilization of a multivariate load time sequence through a data preprocessing module; the feature embedding module performs linear embedding on the block sequence to construct high-dimensional feature representation; the state space coding module extracts long-range dependency features and generates depth time sequence representation; the decoding prediction module maps the coding features into a preliminary prediction sequence; the time sequence alignment module identifies a leading-lagging relation among multiple variables and aligns a time sequence; the frequency domain optimization module realizes frequency domain component fusion based on adaptive filtering; and the model training optimization module is used for performing training and optimization through a signal attenuation loss function. The method can effectively improve the precision and robustness of long-range multivariable load prediction, and especially has obvious advantages in the aspects of processing complex dependency relationships and dynamic time delay.

The invention discloses a multivariable underground water level time sequence prediction method and device, equipment and a medium, and relates to the technical field of artificial intelligence. Performing feature variable contribution proportion calculation, feature screening and normalization processing on the processed data based on a random forest regression model by using an average impurity reduction method to obtain target data; constructing a prediction model based on a convolutional neural network and a long-short term memory network, and performing structure setting, cross validation, parameter setting and model training on the prediction model to obtain a target prediction model; the method comprises the following steps: inputting target data into a target prediction model to obtain a predicted value, generating an evaluation index based on the predicted value and an actual measured value, optimizing the target prediction model by using the evaluation index, improving the applicability, generalization ability and training speed of the model, optimizing the convergence of the model, and improving the accuracy and stability of multivariable groundwater level time sequence prediction. And the sustainable utilization efficiency of underground water resources is improved.

The invention discloses a power transformer temperature prediction method and system based on multi-feature extraction and fusion, and the method comprises the steps: carrying out the time-space decoupling standardization and causal emergence clustering of all temperature variables in data, and obtaining a clustering result; extracting a common pattern from each class of the clustering result to obtain an information complemented multivariable time sequence, and performing down-sampling on the information complemented multivariable time sequence to obtain a plurality of subsequences; performing feature extraction on the multivariable time sequence and the sub-sequence after information completion through a dynamic time-frequency domain MLP mechanism to obtain overall features and local features, and fusing the overall features and the local features to obtain time sequence features; acquiring an external environment covariable, inputting the external environment covariable into the self-attention mechanism network to obtain a covariable self-feature, and performing feature fusion on the covariable self-feature and the time sequence feature through a covariable-driven causal convolution fusion mechanism to obtain a fusion feature; and inputting the fusion feature into a dynamic attention feature weighted prediction model for prediction to obtain a prediction result. And the precision and reliability of temperature prediction of the power transformer are obviously improved.

The invention discloses a drainage basin hydrological multivariable prediction method based on a meta-learning TFT architecture, and relates to the technical field of hydrological prediction, and the method comprises the steps: collecting multivariable water quality data, external driving force data and static drainage basin attribute data of a plurality of hydrological observation stations; splicing and processing missing values, and dividing the missing values into a training set, a test set and a verification set which are arranged in time as input; the capture of the model on the long-term dependency relationship is enhanced by enhancing the gating circulation unit; capturing a multi-level dependency relationship in the time sequence by using a multi-scale image attention network; and integrating the meta-learning framework, fusing a plurality of subtasks, and outputting a predicted value. Model fusion enhanced meta learning and improved Temporal Fusion Transformer are combined, a multi-scale map attention network and a dynamic model fusion mechanism are fused, the defects of a hydrological prediction model in the aspects of capturing a long-term dependency relationship and generalization ability in a complex watershed environment are overcome, and the precision and reliability of multivariable hydrological prediction are improved.

The invention relates to the technical field of mine filling control, in particular to a multivariable layered adaptive filling regulation and control method based on reinforcement learning, and the method comprises the steps: constructing a state vector; inputting the state vector into a target reinforcement learning strategy network model to obtain a target action parameter and state value estimation; judging whether an expected standard is met or not based on the state value estimation, and if yes, adjusting a filling control parameter of the filling system based on the target action parameter; in response to completion of adjustment of the filling control parameters of the filling system, state characteristic functions corresponding to the filling system in the current control period are constructed, and a multi-dimensional reward function is constructed based on the state characteristic functions; and judging whether to trigger strategy update based on the control cycle number and the current accumulated reward value, and if so, updating the target reinforcement learning strategy network based on the multi-dimensional reward function. According to the invention, adaptive coordination optimization of a multivariable target can be realized, and the precision and stability of filling control are effectively improved.

The invention provides an automatic data management method and system based on a multi-modal large model, and the method comprises the steps: collecting multi-source heterogeneous industrial data, and carrying out the standardization processing, and forming standardized multivariable time series data; constructing a process knowledge base, and performing semantic embedding coding on a process knowledge text and storing the process knowledge text; constructing and finely tuning a KTSF multi-modal large model, and fusing process knowledge semantics and multivariable time sequence data through a cross-modal attention mechanism to generate joint semantic representation; based on prediction of a KTSF multi-mode large model, outputting a residual error with actual data, and dynamically identifying abnormal data; performing attribution analysis; based on an attribution result, calling a KTSF multi-mode large model to generate a repair value, and performing intelligent correction on the abnormal data; the design quality evaluation and feedback learning module is used for calculating a data quality score and driving incremental updating of the model; and the design rule self-learning module is used for automatically extracting the governance rule through clustering analysis and updating the knowledge base.

The invention belongs to the technical field of intelligent manufacturing, and particularly relates to a manufacturing quality prediction method and system based on a multi-mode sequential network and application, and the method comprises the steps: carrying out the preprocessing of the sequential data of a process manufacturing production line, obtaining a sample set, and carrying out the sequential division into a training set, a verification set and a test set; on the basis of the sample set, key features are enhanced through a frequency domain enhanced channel attention mechanism, a multi-period mode of a time sequence dependence and period sensing module is captured in combination with a multi-layer expansion convolutional network structure, and a multi-mode time sequence network model is constructed; and sequentially carrying out training set training, verification set parameter adjustment optimization and test set performance verification on the multi-modal sequential network model, and outputting a prediction result. According to the method, the deep dynamic association among the multivariable time series data can be mined, the accuracy and robustness of manufacturing quality prediction are improved, and an efficient and reliable technical scheme and an implementation path are provided for process industry quality control and intelligent optimization.

The invention discloses a corn quality distribution rule analysis method, relates to the technical field of corn quality analysis, and aims to solve the problems of inaccurate corn quality data acquisition and unclear corn quality distribution. According to the method, by combining a univariate analysis method and a multivariate analysis method, the internal rule of the abnormal index data can be comprehensively and deeply mined, the abnormal distribution rule decision is made according to the distribution mode of the abnormal index data, and the potential reason of the corn quality index data abnormity can be scientifically deduced; the standard sample collection data, the abnormal index data and the abnormal distribution rule decision result are integrated and displayed on the display terminal, information comprehensiveness is achieved, the standard sample collection data are compared with the standard index data divided in detail, the data deviation of each specific index can be accurately recognized, and the accuracy of the data deviation is improved. Through comprehensive and meticulous preprocessing of the target index data, powerful support is provided for follow-up study of corn quality distribution rules by using a statistical analysis method.

The invention discloses an industrial Internet of Things system anomaly detection method based on multivariable time series data. The method comprises the following steps: S1, dividing original multivariable time series data into an original training set and an original test set; s2, performing normalization processing on the original training set and the original test set, and performing time window segmentation to obtain a new training set and a new test set; s3, a GTVform model is constructed; s4, the GTVformer model trains the data in the new training set in two stages, and the two Transformer models are trained in a confrontation mode to reconstruct normal input window data; and S5, predicting the data in the new test set by using the trained GTVformer model to obtain a marking result. By combining the time-variable attention mechanism and the adversarial training method, the accuracy of multivariable time series data anomaly detection is remarkably improved, and meanwhile, the robustness of the model is enhanced.

The invention relates to a multivariable multi-step air conditioner load prediction model based on time sequence convolution and a double attention mechanism, and belongs to the technical field of air conditioner load prediction. The model adopts a parallel encoding structure, in an encoder, a time sequence convolution module is responsible for modeling local dependence and long-term trend in a time dimension, and a double attention mechanism module is used for modeling a dynamic dependence structure among multiple variables and a coupling relation between the variables and a target load from a variable dimension. The two structures are respectively subjected to feature extraction from a time dimension and a variable dimension, and are complementary to each other. In a decoder, a decoding module with a memory ability and a teacher mandatory strategy is designed, and continuous prediction from a historical multivariable sequence to a future target load is realized.

The invention belongs to the technical field of information, and discloses an IT performance index early warning method based on time sequence analysis, and the method comprises the steps: S1, inputting historical data; s2, risk detection; and S3, outputting early warning. Through fusion of statistical learning and machine learning technologies, accurate identification and early warning of three scenes of slow degradation of IT performance indexes, mode switching and capacity bottleneck are realized. The method specifically aims to improve the accuracy of performance index trend identification through a variable-point enhanced segmentation regression algorithm, and help an I T manager to more accurately evaluate the influence and trend of performance changes. Designing a cycle-adaptive double-sample hypothesis testing mechanism, and detecting and identifying switching of different load modes, especially significant changes caused by special or emergency events; and constructing a multivariable autoregressive VAR model, establishing a relation model of performance indexes and capacity indexes, evaluating capacity bottleneck conditions, and performing early warning in time.

The invention provides a short-term load prediction method, which comprises the following steps: collecting multivariable data of a power system, and preprocessing the collected multivariable data; wherein the multivariable data at least comprises power load data, weather data, holiday and festival information and date data; inputting the preprocessed multivariable data into a pre-trained prediction model to obtain a corresponding prediction result; wherein the prediction model is a model combining a time convolution network and a long short-term memory network, hyper-parameter optimization is carried out through a preset search algorithm, training is carried out through historical multivariable data, and a plurality of intrinsic mode component prediction results are combined to be output as the model; and determining the short-term load corresponding to the power system according to the prediction result. According to the method, the nonlinear features in the power load data are effectively extracted, more accurate load prediction is realized through advantage complementation of multiple models, and a solid guarantee is provided for stable operation of a power system.

The invention relates to the technical field of phosphoric acid production control, solves the problem that in the prior art, accurate control over the whole process of phosphoric acid production cannot be achieved, and provides a multivariable model prediction control method and system for the whole process of phosphoric acid production. The method comprises the following steps: acquiring key process data of a plurality of process units in the whole phosphoric acid production process in real time; preprocessing the key process data to obtain real-time standardized data; establishing a soft measurement model according to a pre-collected historical standardized data set; inputting the real-time standardized data into the soft measurement model to obtain a soft measurement predicted value; constructing a multivariable dynamic model according to the historical standardized data set and the soft measurement predicted value; according to the multivariable dynamic model, in combination with a model prediction control algorithm, obtaining target set values of a plurality of preset operation variables; and according to the target set value, realizing control of the whole process of phosphoric acid production. According to the invention, accurate control of the whole process of phosphoric acid production can be realized.

The invention discloses a multivariable predictive control energy consumption adjustment method and system, and belongs to the technical field of automatic control, and the method comprises the steps: collecting a parameter adjustment log in real time through an edge node, carrying out the intervention behavior recognition, and verifying the validity, so as to detect a manual parameter adjustment event; once an artificial parameter adjustment event is detected, multivariable data before and after intervention are extracted, and parameters of the local prediction model are dynamically corrected; performing rehearsal intervention based on the manual intervention parameters, generating space-time coupling constraints, and constructing a hybrid neural network to generate an energy consumption prediction trajectory; dividing types according to operator behavior modes, fusing prediction data to generate a comprehensive state vector, adjusting a reward function, focusing a sensitive variable, and generating a control instruction; and acquiring an actual energy consumption value in real time, comparing the actual energy consumption value with an energy consumption prediction track, calculating an energy consumption deviation, performing secondary optimization, positioning an error root cause through multi-scale decomposition in combination with a semantic tag and a knowledge graph, and performing layered compensation.