Real-time high-precision radiation environment monitor based on autoregressive encoder

Abstract

The application discloses a kind of real-time high-precision radiation environment monitoring instrument based on autoregressive encoder, including data acquisition module and host computer;The data acquisition module includes gamma dose rate sensor, temperature sensor and humidity sensor;Data acquisition module and host computer are sequentially connected.The host computer includes data processing module, real-time prediction module, data storage unit and result display module.The real-time high-precision radiation environment monitoring instrument based on the regression encoder disclosed in the application has the advantages of strong real-time, high prediction accuracy.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of environmental monitoring, and particularly relates to a real-time high-precision radiation environmental monitor based on an autoregressive encoder. BACKGROUND

[0002] Radiation environmental monitoring has important application value in the fields of nuclear energy management, environmental protection, public safety, etc. Accurate prediction of radiation signal changes not only provides an important basis for environmental protection, but also has crucial significance for nuclear accident warning and disaster emergency. However, the prediction of radiation signals faces many challenges, mainly from the following aspects:

[0003] Firstly, the sources of radiation signals are diverse, including natural radiation, industrial radiation, and human activities, etc. The intensity and distribution of these signals are affected by many factors such as meteorological conditions, geographical location, etc., so they have high complexity and uncertainty. Secondly, the existing radiation signal collection points are relatively sparse, resulting in limited amount of historical data, and traditional prediction methods are difficult to fully utilize these limited data for efficient prediction. In addition, the radiation environment has high dynamic change characteristics, which requires the prediction system not only to respond in real time, but also to effectively identify complex change trends and sudden events.

[0004] Currently, traditional radiation signal prediction methods mainly rely on statistical analysis and empirical models, which have poor effects in dealing with nonlinear and high-dimensional data, and are difficult to meet the requirements of real-time and high precision. With the development of deep learning technology, more and more intelligent algorithms are applied to radiation signal prediction. However, most of the existing deep learning models are difficult to effectively capture the complex spatiotemporal features in the radiation signal under the condition of limited data, and cannot provide accurate enough prediction results under the requirement of high real-time. Therefore, the core problem of radiation signal prediction is how to accurately extract effective features in the radiation signal under the condition of limited data, and provide high-precision and real-time prediction results.

[0005] In order to overcome these technical bottlenecks, the present application innovatively introduces a framework based on autoregressive encoder, combined with multi-scale feature extraction mechanism, which can not only maintain high prediction accuracy when dealing with complex environmental data, but also has strong real-time and robustness. Specifically, the technical innovation of the present application lies in the use of autoregressive mechanism to dynamically optimize and adjust the prediction results at each step, further improving the accuracy of prediction. In addition, the present application can extract long-term trends and periodic features of radiation signals at multiple levels through multi-scale processing technology, effectively suppressing short-term noise and interference, thereby improving the overall stability and accuracy of the model.

[0006] The embodiment of the present application can significantly improve the accuracy and timeliness of radiation signal prediction, provide a new technical path for radiation environment monitoring, and has important theoretical value and practical application prospect. SUMMARY

[0007] In order to overcome the difficulty of accurately extracting effective features in the radiation signal under limited data conditions and providing high-precision and real-time prediction results, the purpose of the present application is to provide a real-time high-precision radiation environment monitor based on an autoregressive encoder for radiation environment monitoring, which can significantly improve the accuracy and timeliness of radiation signal prediction.

[0008] The technical scheme adopted by the present application to solve its technical problems is: a real-time high-precision radiation environment monitor based on an autoregressive encoder, comprising a data acquisition module and an upper computer; the data acquisition module comprises a gamma dose rate sensor, a temperature sensor and a humidity sensor, which are respectively used to acquire radiation signal data, temperature data and humidity data in the environment in real time; the upper computer comprises a data processing module, a real-time prediction module and a data storage unit;

[0009] The data processing module is used for standardizing the radiation signal, temperature and humidity data in the data acquisition module;

[0010] The real-time prediction module realizes real-time prediction through analysis of historical environment data and real-time environment data by a prediction model, and the environment data comprises radiation signal data, temperature data and humidity data; and a multi-step optimization and correction mechanism is introduced: the prediction model not only uses historical environment data for trend prediction, but also uses the prediction result of each step as input to correct and optimize the subsequent prediction result;

[0011] The data storage unit is used for storing historical environment data, real-time environment data and prediction results generated by the real-time prediction module.

[0012] Further, the data processing module adopts the following process to complete:

[0013] In the data processing module, the radiation signal, temperature and humidity data collected by the data acquisition module need to be standardized to eliminate the influence of different dimensions and orders of magnitude on the subsequent processing process. The formula of the standardization process is as follows:

[0014] For each dimension of data x i , that is, radiation signal, temperature data or humidity data, the standardization process is:

[0015]

[0016] Where x t represents the data after standardization, xi Let represent the i-th data point, and μ be the mean of the dataset, calculated using the following formula:

[0017]

[0018] σ is the standard deviation of the dataset, calculated using the following formula:

[0019]

[0020] Furthermore, the real-time prediction module completes this process as follows:

[0021] The predictive model innovatively incorporates an autoregressive framework, using historical environmental data for multi-step predictions. By continuously feeding back and optimizing the prediction results at each step, the accuracy and stability of the predictions are progressively improved. The prediction at each time step t is expressed as:

[0022]

[0023] Where n represents the length of historical environmental data, x represents the prediction result at time step t. t-1 ,x t-2 ,…,x t-n Given a historical environmental data sequence, f(·) is the prediction function of the prediction model. The results of each prediction are then analyzed. As new input, the system can progressively correct and optimize subsequent predictions:

[0024]

[0025] Where k represents the length of the future prediction.

[0026] Furthermore, the real-time prediction module also includes a seasonal decomposition module to extract long-term trends and periodic features, which can better handle complex and variable environmental data, improve prediction accuracy and stability, and is especially suitable for real-time monitoring scenarios with strong seasonal fluctuations and noise interference, such as radiation environments; this is accomplished through the following process:

[0027] A multi-scale decomposition method is employed to decompose the data into multiple frequencies to accommodate different periodic characteristics. Specifically, the Empirical Mode Decomposition (EMD) method is used to decompose the data layer by layer from low frequency (long-term trends) to high frequency (short-term fluctuations):

[0028]

[0029] Where, x t The data has been standardized; T tThe long-term trend component represents the long-term trend of data change; S t,i As the i-th layer of seasonal fluctuation components, it can capture periodic fluctuations at different time scales; R t The residual component represents noise or short-term fluctuations that cannot be explained by trends and seasonal fluctuations.

[0030] It also includes: the introduction of an adaptive periodicity detection mechanism; the specific periodicity detection is implemented through the following process:

[0031] Frequency Spectrum = FFT(T) t )

[0032] Then, the seasonal period length p is dynamically determined by the frequency component with the highest energy in the spectrum.

[0033]

[0034] Where P is the set of all possible period lengths, FFT p This represents the amplitude corresponding to a specific period in the spectrum. The periodic component S t,i Reconstructed from the detected period length P and the spectral analysis results of the original signal:

[0035]

[0036] Where A and B are amplitude coefficients obtained through spectral analysis.

[0037] It also includes: introducing a nonlinear noise suppression algorithm, combining adaptive filters and Kalman filtering techniques to model and suppress high-frequency noise components; specifically:

[0038] By introducing filters into the high-frequency components, the residual R is improved. t Make corrections and optimizations:

[0039] R′ t =Kalman(R) t ,Q,R)

[0040] Here, Q and R represent the covariance matrices of process noise and observation noise, respectively. The Kalman filter can adaptively adjust the residual components based on historical data and noise characteristics.

[0041] By recombining the components, we obtain the reconstructed x. t ′ :

[0042]

[0043] Furthermore, it also includes: introducing a GQA mechanism, namely Grouped Query Attention, which enables fine-grained attention allocation across different features and time steps when processing complex time-series data, improving prediction accuracy and enhancing the model's adaptability and robustness; the GQA mechanism groups queries, with each group representing a data type or feature at a time scale. The reconstructed x t ′ The input is fed into a linear transformation layer, which maps it to a high-dimensional feature space; the specific formula is as follows:

[0044] Q t,g =W Q ·x′ t +b Q

[0045] K t,g =W K ·x′ t +b K

[0046] Among them, W Q and W K These are the query weight matrix and the key weight matrix, respectively; b Q and b K These are the bias vectors for the query and the key, respectively; Q t,g K is the query vector in the g-th group at time step t; t,g Q is the key vector in the g-th group at time step t; i,g K is the query vector in the g-th group at the i-th time step; i,g It is the key vector in the g-th group at the i-th time step; T represents the total number of time steps; the attention calculation in GQA is expressed as:

[0047]

[0048] α t,g Is it a query Q? t,g The weight in the g-th group; score(·) is the similarity between the query vector and the key vector, calculated using the dot product method.

[0049] It also includes the introduction of an adaptive adjustment mechanism; dynamic adjustments are made using the following formula:

[0050] Q′ t,g =Linear(x t,g +λ·H t,g )

[0051] Where, x t,g H represents the input features of the g-th group at the current time step y; t,g Historical information x from previous predictions t ′λ represents the predictive model's understanding of this set of features; λ is a dynamic weight adjustment factor that automatically adjusts based on the influence of historical environmental data; Q t ′ ,g This is the updated query vector. Based on Q... t ′ ,g Recalculate the attention to obtain α. t ′ ,g Based on attention weight α t ′ ,g For the reconstructed time series x t ′ Perform weighted combination to generate prediction results The formula is:

[0052]

[0053] Furthermore, the data storage unit is used to calculate the prediction error by storing historical environmental data and prediction results generated by the real-time prediction module, in order to dynamically optimize the prediction model parameters; specifically:

[0054] 4.1 Historical environment data storage

[0055] The data storage unit first stores historical environmental data and corresponding real-time environmental data, as well as the prediction results at each time step; this data is used to provide input for training and prediction of the prediction model. For each data point, the system records its original value and corresponding predicted value.

[0056] 4.2 MSE Computation and Storage

[0057] Calculate the error MSE between the actual and predicted values ​​of historical environmental data:

[0058]

[0059] Where n is the number of data points stored; x t This refers to historical environmental data, that is, data that has undergone standardization processing; The calculated MSE (Mean Sequence of Predictions) is used as a feedback signal to the prediction model for parameter updates and optimization. Furthermore, the host computer also includes a results display module to provide various real-time display interfaces, including graphical trend analysis, anomaly alarm prompts, and visualized prediction data reports, facilitating user viewing and management of monitoring data.

[0060] The results display module graphically presents real-time environmental data and prediction results, providing time-series charts of multi-dimensional data, including radiation signals, temperature data, and humidity data. Predicted and actual values ​​for each data type are indicated with different colors or graphics to help users intuitively identify trend changes and error fluctuations. On each monitoring interface, users can see the real-time trends of environmental data and compare them with prediction results; the data types are radiation signals, temperature data, and humidity data.

[0061] The beneficial effects of this invention are: 1) It innovatively introduces the design of an autoregressive framework, which can use historical environmental data to make multi-step predictions, and gradually improve the accuracy and stability of predictions by continuously feeding back and optimizing the prediction results of each step.

[0062] 2) The innovative introduction of seasonal decomposition function to extract long-term trends and periodic features can better cope with complex and variable environmental data, improve prediction accuracy and stability, and is especially suitable for real-time monitoring scenarios with strong seasonal fluctuations and noise interference, such as radiation environment.

[0063] 3) An innovative grouped query attention mechanism is introduced, which can perform fine attention allocation between different features and time steps when processing complex time series data, thereby improving prediction accuracy and enhancing the adaptability and robustness of the model.

[0064] 4) This predictor can significantly improve the accuracy and timeliness of radiation signal prediction, providing a brand-new technical approach for radiation environment monitoring, and has important theoretical value and practical application prospects. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a functional structure diagram of a real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder proposed in this invention. Detailed Implementation

[0067] The present invention will now be described in detail with reference to the accompanying drawings. Unless otherwise specified, the features of the following embodiments and implementations can be combined with each other.

[0068] Example 1

[0069] Reference Figure 1A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder is disclosed, comprising a data acquisition module 4 and a host computer. The data acquisition module 4 includes a gamma dose rate sensor 1, a temperature sensor 2, and a humidity sensor 3, used to collect radiation signal data, temperature data, and humidity data from the environment, respectively. These data form the basis for subsequent processing and environmental monitoring. The data acquisition module 4 and the host computer are connected sequentially. The host computer includes a data processing module 5, a real-time prediction module 6, a data storage unit 7, and a result display module 8.

[0070] Data processing module 5 is responsible for preprocessing the radiation signals, temperature and humidity data in the data acquisition module to ensure the stability and reliability of data acquisition.

[0071] In data processing module 5, the radiation signals, temperature, and humidity data acquired in data acquisition module 4 need to be standardized to eliminate the influence of different dimensions and magnitudes on subsequent processing. The formula for the standardization process is as follows:

[0072] For each dimension of data x i (For example, radiation signals, temperature, or humidity), the standardization process is as follows:

[0073]

[0074] Where x represents the data after standardization, x i Let represent the i-th data point, and μ be the mean of the dataset, calculated using the following formula:

[0075]

[0076] σ is the standard deviation of the dataset, calculated using the following formula:

[0077]

[0078] Through the standardization process described above, the data will be transformed into a standard normal distribution with a mean of 0 and a standard deviation of 1. This processing step effectively avoids the adverse effects of differences in feature scales on the model training process, ensuring equal weight for all types of data in the model, thereby improving prediction accuracy and model training efficiency.

[0079] 2. Real-time prediction module 6 is completed using the following process:

[0080] 2.1 Core Autoregressive Framework and Multi-Step Optimization Mechanism:

[0081] The core innovation of the real-time prediction module 6 is the design of an autoregressive framework, namely the prediction model. This model utilizes historical and real-time environmental data for multi-step predictions, and continuously improves the accuracy and stability of the predictions by providing feedback and optimizing the results at each step. The prediction at each time step t can be expressed as:

[0082]

[0083] Where n represents the length of historical environmental data, x represents the prediction result at time step t. t-1 ,x t-2 ,…,x t-n Given a historical environmental data sequence, f(·) is the prediction function of the prediction model. The results of each prediction are then analyzed. As new input, subsequent predictions are progressively corrected and optimized:

[0084]

[0085] Here, k represents the length of the future prediction. This optimization mechanism based on an autoregressive framework ensures that each prediction result can be used to adjust future predictions, reducing error accumulation and improving accuracy.

[0086] 2.2 Seasonal decomposition function: Extracting long-term trends and cyclical characteristics

[0087] The seasonal decomposition module is a key innovation in the real-time forecasting module 6. Its core objective is to extract and separate the complex temporal components from radiation signals, temperature, and humidity data through multi-level decomposition. This allows for more accurate capture of long-term trends and periodic changes, while effectively removing noise and short-term fluctuations, thus improving forecast accuracy and stability. Traditional seasonal decomposition methods typically split data into three parts: long-term trend, seasonal fluctuations, and residuals. This design, however, employs a complex multi-scale decomposition method and an adaptive forecasting model to address the changing characteristics of environmental data.

[0088] Traditional seasonality decomposition methods typically analyze data based on a fixed time scale. However, in real-world environments, data such as radiation signals, temperature, and humidity may exhibit seasonal variations at multiple frequencies. For example, radiation signals may show annual periodic fluctuations, while temperature and humidity may exhibit strong seasonal fluctuations over shorter periods (such as diurnal variations). Therefore, this module employs a multi-scale decomposition method to decompose the data at multiple frequencies to accommodate different periodic characteristics. Specifically, it uses Empirical Mode Decomposition (EMD) to decompose the data layer by layer from low frequency (long-term trends) to high frequency (short-term fluctuations):

[0089]

[0090] Where, x t The data has been standardized; T t The long-term trend component represents the long-term trend of data change; S t,i As the i-th layer of seasonal fluctuation components, it can capture periodic fluctuations at different time scales; R t The residual component represents noise or short-term fluctuations that cannot be explained by trends and seasonal fluctuations. This multi-scale decomposition not only accurately extracts periodic fluctuations of different frequencies but also provides fine-grained trend modeling, thereby helping predictive models better understand the structure of the data.

[0091] To further enhance adaptability to different data periods, this module introduces an adaptive period detection mechanism. In traditional seasonal decomposition, the period length is usually preset, which cannot flexibly cope with dynamic changes in the period within the data. In this module, period detection is dynamically adaptive, automatically determining the optimal period length based on the characteristics of the data. This mechanism relies on spectral analysis of the data (e.g., through Fast Fourier Transform, FFT) to determine the period by detecting significant frequency components in the data. The specific period detection can be implemented through the following process:

[0092] Frequency Spectrum = FFT(T) t )

[0093] Then, the seasonal period length p is dynamically determined by the frequency component with the highest energy in the spectrum.

[0094]

[0095] Where P is the set of all possible period lengths, FFT p This represents the amplitude corresponding to a specific period in the spectrum. The periodic component S t,i It can be reconstructed from the detected period length P and the spectral analysis results of the original signal:

[0096]

[0097] Where A and B are amplitude coefficients obtained through spectral analysis. Through adaptive period detection, the system can automatically adjust the period of seasonal decomposition according to changes in environmental data, enabling the prediction model to accurately capture potential seasonal patterns in the data.

[0098] During seasonal decomposition, unavoidable noise or short-term fluctuations often occur. Traditional methods typically classify these fluctuations as "residuals," but in some cases, this noise may contain useful information, especially for nonlinear and sudden environmental changes. Therefore, this module introduces a nonlinear noise suppression algorithm, combining adaptive filters and Kalman filtering techniques to accurately model and suppress high-frequency noise components.

[0099] This method introduces a filter into the high-frequency components to improve the residual R. t Corrections and optimizations are performed to avoid the impact of noise interference on the prediction results.

[0100] R′ t =Kalman(R) t ,Q,R)

[0101] Here, Q and R represent the covariance matrices of process noise and observation noise, respectively. The Kalman filter can adaptively adjust the residual components based on historical environmental data and noise characteristics.

[0102] Recombining the components yields the reconstructed x′. t :

[0103]

[0104] Through the above innovative methods, the seasonal decomposition function can better cope with complex and variable environmental data, improve prediction accuracy and stability, and is especially suitable for real-time monitoring scenarios with strong seasonal fluctuations and noise interference, such as radiation environment.

[0105] 2.3 Grouped Query Attention Mechanism

[0106] To further enhance the adaptability of the real-time prediction module 6 to complex environmental data, this module introduces the Grouped Query Attention (GQA) mechanism. As an innovative attention model, it can perform fine-grained attention allocation across different features and time steps when processing complex time-series data, improving prediction accuracy and enhancing the model's adaptability and robustness. In this design, the GQA mechanism groups queries, with each group representing a data type or feature at a specific time scale. The query vector within each group is specifically used to process data related to that feature type, enabling the model to adaptively adjust its attention allocation across different data dimensions. The reconstructed x′ is input to a linear transformation layer, mapping it to a high-dimensional feature space. The specific formula is as follows:

[0107] Q t,g =W Q ·x′ t +bQ

[0108] K t,g =W K ·x′ t +b K

[0109] Among them, W Q and W K These are the query weight matrix and the key weight matrix, respectively; b Q and b K These are the bias vectors for the query and the key, respectively; Q t,g K is the query vector in the g-th group at time step t; t,g It is the key vector in the g-th group at time step t; specifically, the attention calculation of Grouped Query Attention can be expressed as:

[0110]

[0111] α t,g Is it a query Q? t,g The weight in the g-th group; score(·) is the similarity between the query vector and the key vector, calculated using the dot product method.

[0112] Beyond query grouping, this module's innovation lies in its adaptive adjustment mechanism. Based on changes in real-time environmental data, the predictive model automatically determines which features require more attention. Through adaptive learning, the system can adjust the weights of each group at different time steps, flexibly responding to various environmental changes. For example, when a certain dimension (such as temperature) becomes more influential on radiation signals, the system will increase the query weight of that dimension; while at other times, the system will automatically reduce its focus on that dimension.

[0113] This adaptive mechanism is dynamically adjusted using the following formula:

[0114] Q′ t,g =Linear(x t,g +λ·H t,g )

[0115] Where, x t,h For the current time step t, the input features of the g-th group (such as temperature, humidity, etc.); H t,g The historical information from previous predictions represents the predictive model's understanding of this set of features; λ is a dynamic weight adjustment factor that automatically adjusts based on the influence of historical environmental data; Q′ t,g This is the updated query vector. Based on Q′ t,g Recalculate the attention to obtain α′ t,g Based on attention weight α′t,g For the reconstructed time series x′ t Perform weighted combination to generate prediction results The formula is:

[0116]

[0117] Through this mechanism, the prediction model can automatically adjust the query vectors of each group according to changes in feature importance, ensuring accurate capture of key features. By introducing Grouped Query Attention, this module can perform fine-grained attention allocation among different features and time steps when processing complex time-series data, improving prediction accuracy and enhancing the adaptability and robustness of the prediction model.

[0118] 3. Data storage unit 7 is completed using the following process:

[0119] Data storage unit 7 plays a crucial role in the entire system, responsible for storing historical environmental data, real-time environmental data, and real-time prediction results, and providing data support for the autoregressive framework. This unit not only stores and maintains data but also provides the foundation for the optimization process of autoregressive prediction. By storing historical environmental data and prediction results, it calculates and feeds back the mean squared error (MSE), thereby optimizing the prediction model and improving its accuracy. Data storage unit 7 ensures the stability and efficiency of the prediction model and is a key component of the autoregressive framework and prediction optimization.

[0120] 3.1 Historical environment data storage

[0121] Data storage unit 7 first stores all historical environmental data and corresponding real-time environmental data (such as radiation signals, temperature, humidity, etc.), as well as the prediction results for each time step. This data is used to provide input for training and prediction of the prediction model. For each data point, the system records its original value and corresponding predicted value.

[0122] 3.2 MSE Computation and Storage

[0123] In the autoregressive framework, the error between the actual and predicted values ​​of historical environmental data is crucial to the optimization process. Data storage unit 7 is responsible for calculating and storing the MSE for historical time steps. This error calculation formula is as follows:

[0124]

[0125] Where n is the number of data points stored; x t Historical environmental data; This represents the historical prediction results. The calculated MSE will be used as a feedback signal to the prediction model for parameter updates and optimization within the autoregressive framework. Specifically, the storage unit will store the MSE value corresponding to each time step t, so that the performance of the prediction model can be evaluated and adjusted subsequently.

[0126] 4. The host computer also includes: a result display module 8.

[0127] The results display module 8 graphically presents real-time environmental data and prediction results, providing time-series charts of multi-dimensional data including radiation signals, temperature, and humidity. Predicted and actual values ​​for each data type (such as radiation signals, temperature, and humidity) are indicated by different colors or graphics, helping users intuitively identify trend changes and error fluctuations. On each monitoring interface, users can see the real-time trends of environmental data and compare them with prediction results. This real-time display function helps users promptly grasp the dynamic changes in the radiation environment, providing invaluable information, especially during environmental monitoring, early warning, and response.

[0128] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder, characterized in that: It includes a data acquisition module and a host computer; the data acquisition module includes a gamma dose rate sensor, a temperature sensor and a humidity sensor, which are used to collect radiation signal data, temperature data and humidity data in the environment in real time, respectively; the host computer includes a data processing module, a real-time prediction module and a data storage unit; The data processing module is used to standardize the radiation signal, temperature and humidity data in the data acquisition module. The real-time prediction module achieves real-time prediction by analyzing historical and real-time environmental data through a prediction model. The environmental data includes radiation signal data, temperature data, and humidity data. A multi-step optimization and correction mechanism is introduced: the prediction model not only uses historical environmental data for trend prediction, but also uses the prediction results of each step as input to correct and optimize subsequent prediction results. The real-time forecasting module also includes a seasonal decomposition module, which is completed using the following process: Using the empirical mode decomposition method, the data is decomposed layer by layer from low frequency to high frequency: ； in, This is data that has undergone standardization. The long-term trend component represents the long-term trend of data change. For the first The seasonal fluctuation component can capture periodic fluctuations at different time scales; The residual component represents noise or short-term fluctuations that cannot be explained by trends and seasonal fluctuations. It also includes: the introduction of an adaptive periodicity detection mechanism; the specific periodicity detection is implemented through the following process: ； Then, the length of the seasonal cycle is dynamically determined by the frequency component with the highest energy in the spectrum. : ； in, For the set of all possible seasonal cycle lengths, The amplitude corresponding to a certain period in the spectrum; seasonal fluctuation component Set of seasonal cycle lengths obtained from detection Reconstructed from the spectral analysis results of the original signal: ； in, and These are the amplitude coefficients obtained through spectral analysis; It also includes: introducing a nonlinear noise suppression algorithm, combining adaptive filters and Kalman filtering techniques to model and suppress high-frequency noise components; specifically: By introducing filters into the high-frequency components, the residuals are... Make corrections and optimizations: ； in, and These represent the covariance matrices of process noise and observation noise, respectively. The Kalman filter can adaptively adjust the residual components based on historical data and noise characteristics. The components are then recombined to obtain the reconstructed time series. : ； The data storage unit is used to store historical environmental data, real-time environmental data, and prediction results generated by the real-time prediction module.

2. The real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder according to claim 1, characterized in that: The data processing module completes the task using the following process: In the data processing module, the radiation signals, temperature, and humidity data acquired by the data acquisition module need to be standardized; the formula for the standardization process is as follows: For each dimension of data This refers to radiation signals, temperature data, or humidity data, and its standardization process is as follows: ； in This indicates data that has undergone standardization. Indicates the first Data points, The mean of the dataset is calculated using the following formula: ； The standard deviation of the dataset is calculated using the following formula: 。 3. A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder according to claim 1, characterized in that: The real-time prediction module is completed using the following process: The predictive model utilizes historical environmental data for multi-step predictions, and continuously improves the accuracy and stability of predictions by providing feedback and optimizing the results of each step. At each time step... The prediction is expressed as: ； in, Indicates the length of historical environmental data. Indicates at time step The prediction results For historical environmental data sequences, The prediction function of the prediction model; by using the results of each prediction As new input, the system can progressively correct and optimize subsequent predictions: ； in, Indicates the length of the future forecast.

4. A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder according to claim 1, characterized in that: Also includes: Introducing the GQA mechanism; the GQA mechanism groups queries, with each group representing a data type or a feature at a time scale. Reconstructed time series It is input into a linear transformation layer, which maps it to a high-dimensional feature space; The specific formula is as follows: ； ； in, and These are the query weight matrix and the key weight matrix, respectively. and These are the bias vector for the query and the bias vector for the key, respectively. It is the first At the time step The query vector within the group; It is the first At the time step Key vectors within a group; It is the first At the time step The query vector within the group; It is the first At the time step The key vector within the group; T represents the total time steps; GQA attention calculation is represented as: ； It is a query In the Weights within the group; It is the similarity between the query vector and the key vector, calculated using the dot product method; It also includes the introduction of an adaptive adjustment mechanism; dynamic adjustments are made using the following formula: ； in, For the current time step For the first Group input features; Reconstructed time series based on previous predictions This indicates the degree to which the predictive model understands the set of features; It is a dynamic weighting adjustment factor that automatically adjusts based on the impact of historical environmental data. For the updated query vector; based on Recalculate attention to obtain Based on attention weights For the reconstructed time series Perform weighted combination to generate prediction results Its formula is: 。 5. A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder according to claim 1, characterized in that: The data storage unit is used to store historical environmental data and prediction results generated by the real-time prediction module, calculate prediction errors, and dynamically optimize prediction model parameters; specifically: 4.1 Historical environment data storage The data storage unit first stores historical environmental data and corresponding real-time environmental data, as well as prediction results at each time step; this data is used to provide training and prediction input for the prediction model; for each data point, the system records its original value and corresponding predicted value; 4.2 MSE Calculation and Storage Calculate the error between the actual and predicted values ​​of historical environmental data. : ； in, The number of data points to be stored; This refers to historical environmental data, that is, data that has undergone standardization processing; The calculated MSE is the result of historical predictions; it will be used as a feedback signal to be passed to the prediction model for parameter updates and optimization.

6. A real-time high-precision radiation environment monitoring instrument based on an autoregressive encoder according to claim 1, characterized in that: The host computer also includes a result display module, which provides various forms of real-time display interfaces, including graphical trend analysis, abnormal alarm prompts, and visualized predictive data reports, making it convenient for users to view and manage monitoring data; The results display module presents environmental data and prediction results in a graphical manner, providing time series charts of multi-dimensional data, including radiation signals, temperature data, and humidity data. The predicted and actual values ​​for each data type are marked with different colors or graphics to help users intuitively identify trend changes and error fluctuations. On each monitoring interface, users can see the changing trends of environmental data in real time, as well as the comparison with the prediction results. Data types include radiation signals, temperature data, and humidity data.

Images