Frequency control method and system for power grid tool management RFID system

Abstract

The invention discloses a frequency control method and system for a power grid tool management RFID system, relates to the technical field of power equipment management, and realizes accurate prediction and dynamic regulation and control of the working frequency of a reader-writer in the RFID system by introducing a Transform model and a self-adaptive control algorithm. The optimal working frequency of the reader-writer can be effectively predicted by comprehensively analyzing environmental factors, tag characteristics, reader-writer characteristics and time sequence data, and the stability and precision of frequency adjustment are ensured by using methods such as a PID controller and a low-pass filter. According to the control method and system, the communication quality and stability of the RFID system are improved, the adaptability of the system under different environment conditions is enhanced, and long-time stable operation is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment management, and in particular to a frequency control method and system for a power grid tool management RFID system. Background Art

[0002] RFID technology uses radio waves to identify and track objects. Its application areas include logistics management, supply chain tracking, warehouse management, and intelligent manufacturing. The core components of the RFID system include tags, readers, and middleware systems. Tags are attached to objects and store the unique identification information of the objects, while readers communicate with tags through wireless signals to read and write data. RFID systems face a series of technical challenges in practical applications, especially in the management of power grid tools. First, environmental factors (such as temperature, humidity, interference from metal materials, etc.) have a significant impact on the communication quality and performance of RFID systems. Second, the characteristics of tags (such as tag type, aging, etc.) and the characteristics of readers (such as reader power, location, etc.) also affect the recognition accuracy of the system. In addition, the RFID system is usually a time series system, in which the communication between readers and tags changes over time. How to effectively process these time series data to improve the performance of the system is a key issue. At present, although there have been some studies and solutions that attempt to improve the performance of RFID systems by optimizing hardware design and improving communication protocols, these methods are usually difficult to cope with frequency fluctuations and system instability in complex environments. In addition, existing methods often lack the full utilization of time series data, resulting in poor performance of the system in dynamic environments. Summary of the invention

[0003] In view of the above problems, the present invention proposes a frequency control method and system for a power grid tool management RFID system, so that in the power grid tool management application, the operating frequency of the reader / writer can be adjusted in real time under complex environmental conditions, ensuring that the RFID system can maintain good performance and stability in various application scenarios.

[0004] A technical solution of the present invention is: A method for controlling the frequency of an RFID system for managing power grid tools, the method comprising: Collect historical data of RFID system; Performing data fusion on the historical data of the RFID system to obtain fused data; Based on the fused data, respectively, an embedding vector and a position encoding vector of the fused data are obtained through an embedding layer and a position encoding vector of a Transformer model, and input data of the Transformer model is constructed based on the embedding vector and the position encoding vector; Training a Transformer model based on input data of the Transformer model; The trained Transformer model is used to predict the operating frequency of the RFID reader, and the prediction result is used as the target frequency to regulate and constrain the operating frequency of the RFID reader.

[0005] Furthermore, the historical data of the RFID system includes: environmental factor data, tag characteristic data, reader / writer characteristic data and time series characteristic data; The environmental factor data include: temperature ,humidity , metal material distance ; The tag characteristic data includes: tag type , aging degree ; The reader characteristic data includes: reader power , reader position ; The time series characteristic data includes: frequency data , timestamp ; The step of fusing the historical data of the RFID system to obtain fused data includes: Preprocessing the historical data of the RFID system; The discrete feature data in the preprocessed data is encoded through one-hot feature encoding; The discrete feature data after feature encoding is merged with the continuous feature data in the preprocessed data to form a unified feature vector; Fusing the time series feature data with the feature vector to obtain fused data; The discrete feature data includes: label type , reader position ; The continuous characteristic data includes: temperature ,humidity , metal material distance , aging degree , reader power , frequency data .

[0006] Furthermore, the discrete feature data in the preprocessed data is encoded by one-hot, including: Select the discrete feature data that needs to be One-Hot encoded; According to the categories of the tag type and the categories of the reader position, a tag type binary vector and a reader position binary vector are created, wherein the length of the vector is equal to the number of feature categories, and each category corresponds to a position in the vector; the categories of the tag type include: passive tags, active tags, and semi-passive tags; the categories of the reader position include: entrance, exit, production line, and warehouse; The label type binary vector refers to a binary vector of length 3, expressed as: ; in, Refers to passive tags, refers to active tags, It refers to semi-passive tags; The reader position binary vector refers to a binary vector with a length of 4, expressed as: ; in, It refers to the entrance. It refers to the exit. It refers to the production line. It refers to the warehouse; The step of fusing the time series feature data with the feature vector to obtain fused data includes: The timestamps of the time series data are synchronized with the feature data in time, and the corresponding feature vector is found for each timestamp; the time series data and the feature vector are fused to obtain the fused data, which is expressed as: X ' (t)=[X(t),F(t), T s (t)] ; in, refers to the fused data. It refers to time The feature vector at time, It refers to time Frequency data at time, It refers to time The timestamp of the moment, It refers to the concatenation of multiple vectors or values.

[0007] Furthermore, obtaining an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and constructing input data of the Transformer model based on the embedding vector and the position encoding vector, includes: Mapping the feature vector of the fused data to a high-dimensional space through the embedding layer to obtain an embedding vector; The position encoder adds the time position information to each time point to obtain a position encoding vector; The embedding vector and the position encoding vector are added together to obtain the input data of the Transformer model.

[0008] Furthermore, the calculation formula of the position encoding vector is: ; ; in, is the position encoding vector, refers to the dimension index of the position encoding, refers to the dimension of the embedding vector.

[0009] Furthermore, the Transformer model calculates the attention weights between input vectors through a multi-head self-attention mechanism at the encoder layer; performs nonlinear transformation on the output of the multi-head self-attention mechanism, adds residual connections after each sub-layer, and performs layer normalization, and finally constructs a deep feature representation by stacking multiple encoder layers; The Transformer model decoder output is mapped to the target prediction space, and the frequency value of the prediction result is output by the linear layer.

[0010] Furthermore, the attention head calculation formula of the multi-head self-attention mechanism is: ; in, is the dot product of the query matrix and the key matrix, is the scaling factor, is the dimension of the key vector, is a matrix of values; The calculation formula of the nonlinear transformation is: ; in, is the input vector, and is the weight matrix, and is the bias term, It refers to the activation function.

[0011] Furthermore, the training of the Transformer model based on the input data of the Transformer model includes: Use mean squared error as the loss function; Use Adam optimizer to update parameters; Perform forward propagation on the training set, calculate the loss, perform backpropagation and parameter update, iteratively train the model until the loss function converges, and test the model on the validation set to calculate the validation loss.

[0012] Furthermore, the method of using the prediction result as the target frequency to regulate and constrain the working frequency of the RFID reader includes: Set the working frequency of the reader / writer according to the prediction results, use the prediction results as the target frequency, define the upper and lower limits of the reader / writer frequency adjustment, and ensure that the frequency adjustment does not exceed the working range of the device; Based on the discrete time system, a dynamic model of the reader's operating frequency is constructed, and the error between the target frequency and the actual frequency is calculated; The PID controller is used to compensate the frequency error, and the gain coefficient of the PID controller is adjusted in real time based on the changes in the system response; The control signal is smoothed by a low-pass filter; the filtered control signal is applied to the system to update the working frequency of the reader; Calculate the fluctuation range and error variance of the reader frequency, evaluate the frequency stability after adjustment, update the system model and controller parameters through adaptive algorithms based on the adjustment results, and dynamically adjust the key parameters of the system model and controller according to the feedback data during operation; Continuously optimize control algorithms through real-time monitoring and feedback mechanisms; The dynamic model expression is: ; in, Refers to the control input, which means that at time The adjustment amount of the reader frequency at all times, It refers to the actual operating frequency; The error calculation formula is: ; in, Refers to the error between the target frequency and the actual frequency, which is used to adjust the control input. is the target frequency, It refers to the actual operating frequency; The output of the PID controller is: ; in, is the proportional gain, is the integral gain, is the differential gain, is the cumulative sum of errors, It refers to the rate of change of error; The smoothing expression is: ; in, is the filtered control input, is the filter coefficient, is the control input, Refers to the filter control input at the previous moment; The update reader's operating frequency is expressed as: ; in, is the actual operating frequency, Refers to the filtered control input.

[0013] Based on the above-mentioned power grid tool management RFID system frequency control method, the present invention also provides a power grid tool management RFID system frequency control system, the system comprising: A data processing module is used to collect historical data of the RFID system, perform data fusion on the historical data of the RFID system to obtain fused data, obtain an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and construct input data of the Transformer model based on the embedding vector and the position encoding vector; A training module, used for training a Transformer model based on input data of the Transformer model; The control module is used to use the trained Transformer model to predict the operating frequency of the RFID reader, and use the prediction result as the target frequency to regulate and constrain the operating frequency of the RFID reader.

[0014] The power grid tool management RFID system frequency control method and system provided by the embodiment of the present invention collects and analyzes the historical data of the RFID system, combines the time series characteristics and environmental characteristics, and accurately predicts and regulates the working frequency of the reader / writer, thereby improving the stability and communication quality of the system. By introducing an adaptive control algorithm, the method can adjust the working frequency of the reader / writer in real time under complex environmental conditions, ensuring that the RFID system can maintain good performance and stability in various application scenarios.

[0015] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0017] Figure 1 A specific flow chart of the frequency control method of the power grid tool management RFID system provided by an embodiment of the present invention is shown.

[0018] Figure 2 A block diagram of a frequency control system of a power grid tool management RFID system provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0019] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0020] RFID technology uses radio waves to identify and track objects. Its application areas include logistics management, supply chain tracking, warehouse management, and smart manufacturing. The core components of the RFID system include tags, readers, and middleware systems. Tags are attached to objects and store the unique identification information of the objects, while readers communicate with tags through wireless signals to read and write data. Existing RFID system frequency control methods are usually difficult to cope with frequency fluctuations and system instability in complex environments. In addition, existing methods often lack full utilization of time series data, resulting in poor performance of the system in dynamic environments.

[0021] Based on the above problems, the technical solution of the present invention collects and analyzes the historical data of the RFID system, combines the time series characteristics and environmental characteristics, and accurately predicts and regulates the working frequency of the reader, thereby improving the stability and communication quality of the system. By introducing an adaptive control algorithm, this method can adjust the working frequency of the reader in real time under complex environmental conditions, ensuring that the RFID system can maintain good performance and stability in various application scenarios.

[0022] The specific implementation includes the following: Example

[0023] refer to Figure 1 , this embodiment provides a method for controlling the frequency of a power grid tool management RFID system, the method comprising: Step S1: collecting historical data of the RFID system, performing data fusion on the historical data of the RFID system to obtain fused data, obtaining an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and constructing input data of the Transformer model based on the embedding vector and the position encoding vector; Step S2: training the Transformer model based on the input data of the Transformer model; Step S3: Use the trained Transformer model to predict the operating frequency of the RFID reader, and use the prediction result as the target frequency to regulate and constrain the operating frequency of the RFID reader.

[0024] In step S1, the historical data of the RFID system includes: environmental factor data, tag characteristic data, reader characteristic data and time series characteristic data; the environmental factor data includes: temperature ,humidity , metal material distance ; The label characteristic data includes: label type , aging degree ; The reader characteristic data includes: reader power , reader position ; The time series characteristic data includes: frequency data , timestamp .

[0025] In the specific implementation, it is necessary to perform data preprocessing operations on the historical data to obtain preprocessed data, and divide it into a training set and a test set. The preprocessing operations include: data cleaning, data standardization, and data normalization; division refers to dividing the preprocessed data into a training set and a test set in a ratio of 8:2, which are used for training and testing of the Transformer model respectively.

[0026] In the above steps, the historical data of the RFID system is fused to obtain fused data, including: preprocessing the historical data of the RFID system; encoding the discrete feature data in the preprocessed data through one-hot; fusing the feature-encoded discrete feature data with the continuous feature data in the preprocessed data to form a unified feature vector; fusing the time series feature data with the feature vector to obtain fused data, and the feature vector is expressed as: X=[ X one-hot , X cont ] (1); in, refers to the fused feature vector, refers to the encoded discrete feature vector, is a continuous eigenvector, It refers to the concatenation operation of features.

[0027] In the above feature data, the discrete feature data includes: label type , reader position ; Continuous characteristic data, including: temperature ,humidity , metal material distance , aging degree , Reader power , frequency data .

[0028] In this embodiment, the discrete feature data in the preprocessed data is feature encoded through one-hot, including: selecting discrete feature data that needs to be One-Hot encoded; creating a tag type binary vector and a reader position binary vector according to the category of the tag type and the category of the reader position, the length of the vector is equal to the number of feature categories, and each category corresponds to a position in the vector; the category of the tag type includes: passive tags, active tags and semi-passive tags; the category of the reader position includes: entrance, exit, production line and warehouse.

[0029] The label type binary vector refers to a binary vector of length 3, expressed as: (2); in, Refers to passive tags, refers to active tags, It refers to semi-passive tags; The reader position binary vector refers to a binary vector with a length of 4, expressed as: (3); in, It refers to the entrance. It refers to the exit. refers to the production line, It refers to the warehouse.

[0030] The step of fusing the time series feature data with the feature vector to obtain fused data includes: The timestamps of the time series data are synchronized with the feature data in time, and the corresponding feature vector is found for each timestamp; the time series data and the feature vector are fused to obtain the fused data, which is expressed as: X ' (t)=[X(t),F(t), T s (t)] (4); in, refers to the fused data. It refers to time The feature vector at time, It refers to time Frequency data at time, It refers to time The timestamp of the moment, It refers to the concatenation of multiple vectors or values.

[0031] In this step, the embedding vector and the position encoding vector of the fused data are obtained through the embedding layer and the position encoder of the Transformer model respectively based on the fused data, and the input data of the Transformer model is constructed based on the embedding vector and the position encoding vector, including: mapping the feature vector of the fused data to a high-dimensional space through the embedding layer to obtain an embedding vector; adding time position information for each time point through the position encoder to obtain a position encoding vector; and adding the embedding vector and the position encoding vector to obtain the input data of the Transformer model.

[0032] The embedding vector and the position encoding vector are added together. The addition process ensures that the embedding vector of each input contains not only feature information but also temporal position information, thereby helping the model capture temporal dependencies.

[0033] The calculation formula of the position encoding vector is: (5); (6); in, Refers to the position encoding vector, which is used to add time information to the Transformer model that does not have time series information. refers to the dimension index of the position encoding, refers to the dimension of the embedding vector, and Refers to the sine and cosine functions, which are used to generate the periodic components of position encoding. Position encoding is a mechanism used to introduce the position information of elements in the sequence in the Transformer model, because the Transformer itself does not have the ability to process sequence information. In this step, position encoding is to attach the position information to the feature vector at each time point after the input features are mapped to the high-dimensional space through the embedding layer. Because the original embedding vector itself has no time series information, the main purpose of position encoding is to let the model know the position or order of these embedded vectors in the sequence.

[0034] In the present invention, the Transformer model calculates the attention weights between input vectors through a multi-head self-attention mechanism at the encoder layer; performs a nonlinear transformation on the output of the multi-head self-attention mechanism, adds a residual connection after each sublayer, and performs layer normalization, and finally constructs a deep feature representation by stacking multiple encoder layers; the Transformer model decoder output is mapped to the target prediction space, and the frequency value of the prediction result is output by the linear layer.

[0035] Among them, the attention head calculation formula of the multi-head self-attention mechanism is: (7); in, It refers to the dot product of the query matrix and the key matrix, which is used to calculate the similarity score. is the scaling factor, refers to the dimension of the key vector, which is used to prevent the dot product value from being too large. It refers to the value matrix, which is weighted and summed by the attention weights to generate the output.

[0036] The calculation formula for nonlinear transformation is: (8); in, is the input vector, and Refers to the weight matrix, which is used for linear transformation, and Refers to the bias term, which is used to adjust the result of linear transformation. It refers to the activation function, which adds nonlinear characteristics.

[0037] In step S2, the Transformer model is trained based on the input data of the Transformer model, including: using mean square error as the loss function; using Adam optimizer to update parameters; performing forward propagation on the training set, calculating loss, performing back propagation and parameter update, iteratively training the model until the loss function converges, and testing the model on the validation set to calculate the validation loss.

[0038] Specifically, the mean square error is used as the loss function to measure the error between the model prediction value and the true frequency value; the mean square error is expressed as: (9); in, is the total number of data points, is the frequency value predicted by the model, Refers to the real frequency value; the data point refers to the data sample at each moment in the model training process. Specifically, the RFID system collects the frequency information of the reader at different moments, as well as related features (such as tag type, environmental information, etc.). This information constitutes the feature vector at each time point. Each moment corresponds to a specific input data (feature vector and time series information) and a real output value (that is, the real working frequency of the reader). Therefore, a data point refers to a set of input data and the corresponding real frequency value corresponding to each moment t in the time series. The Transformer model is trained with these data points to predict the frequency value at each moment. , and with the real frequency value Make comparisons and calculate errors.

[0039] The Adam optimizer is used to update the parameters, and the parameter update adjustment formula is: (10); in, refers to the updated value of the model parameters during the training process, refers to the current value of the model parameters during training, is the learning rate, is the second-order momentum of the gradient, It refers to the gradient of the loss function with respect to the parameters, Refers to a constant to prevent division by zero errors.

[0040] After training, use the trained Transformer model to predict new time series data, compare the predicted values ​​with the actual frequencies, evaluate the effectiveness of the model in practical applications, and save the trained model parameters if the model performs well.

[0041] In step S3, the prediction result is used as the target frequency to regulate and constrain the working frequency of the RFID reader, including: setting the working frequency of the reader according to the prediction result, using the prediction result as the target frequency, defining the upper and lower limits of the reader frequency adjustment, and ensuring that the frequency adjustment does not exceed the working range of the device; constructing a dynamic model of the reader working frequency based on a discrete time system, and calculating the error between the target frequency and the actual frequency; using a PID controller to compensate for the frequency error, and adjusting the gain coefficient of the PID controller in real time based on changes in the system response; smoothing the control signal through a low-pass filter; applying the filtered control signal to the system to update the working frequency of the reader; calculating the fluctuation range and error variance of the reader frequency, evaluating the frequency stability after adjustment, updating the system model and controller parameters through an adaptive algorithm based on the adjustment results, and dynamically adjusting the key parameters of the system model and controller according to feedback data during operation; continuously optimizing the control algorithm through real-time monitoring and feedback mechanisms; In the above steps, the dynamic model expression is: (11); in, Refers to the control input, which means that at time The adjustment amount of the reader frequency at all times, It refers to the actual operating frequency; The error calculation formula is: (12); in, Refers to the error between the target frequency and the actual frequency, which is used to adjust the control input. is the target frequency, It refers to the actual operating frequency; The output of the PID controller is: (13); in, Refers to the proportional gain, the proportional coefficient of the controller, which is used to amplify the error signal and directly affects the size of the control output. Refers to the integral gain, the integral coefficient of the controller, which is used to accumulate the sum of historical errors to eliminate the cumulative effect of errors. Refers to the differential gain, the differential coefficient of the controller, which is used to respond to the rate of change of the error and help prevent drastic changes in the error. Refers to the cumulative sum of errors, which represents the sum of all errors from the initial moment to the current moment, and is used for integral control. It refers to the rate of change of the error, which represents the derivative of the error in time and is used for differential control.

[0042] The smoothing expression is: (14); in, is the filtered control input, is the filter coefficient, is the control input, Refers to the filter control input at the previous moment.

[0043] The update reader's operating frequency is expressed as: (15); in, is the actual operating frequency, Refers to the filtered control input.

[0044] Calculate the fluctuation range and error variance of the reader frequency, evaluate the frequency stability after adjustment, and based on the adjustment results, update the system model and controller parameters through the adaptive algorithm. According to the feedback data during actual operation, dynamically adjust the key parameters of the system model and controller to improve the accuracy and stability of frequency control. The adaptive algorithm will analyze the error between the target frequency and the actual frequency in real time. By adjusting the controller's gain parameters including the proportional, integral, and differential coefficients in PID control and optimizing the system's dynamic model, the system can adapt to changes in different environmental factors and maintain the best frequency control effect. This adaptive adjustment helps to improve the system's response speed, reduce error fluctuations, and maintain the stability of the system in the long term, so that it can perform better in future frequency control; establish a real-time monitoring and feedback mechanism, and continuously optimize the control algorithm based on the actual system operation through long-term data collection and analysis to ensure that the system can maintain stable communication quality and frequency control effects under various environmental conditions.

[0045] Embodiment 2: refer to Figure 2 Based on the power grid tool management RFID system frequency control method of the above-mentioned embodiment 1, the present invention also provides a power grid tool management RFID system frequency control system, the system comprising: A data processing module is used to collect historical data of the RFID system, perform data fusion on the historical data of the RFID system to obtain fused data, obtain an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and construct input data of the Transformer model based on the embedding vector and the position encoding vector; A training module, used for training a Transformer model based on input data of the Transformer model; The control module is used to use the trained Transformer model to predict the operating frequency of the RFID reader, and use the prediction result as the target frequency to regulate and constrain the operating frequency of the RFID reader.

[0046] The specific implementation method of this embodiment is the same as that of Embodiment 1, which will not be repeated here. Please refer to the description of Embodiment 1 for details.

[0047] By adopting the technical solution of the above embodiment, in the frequency control of the RFID system for power grid tool management, the Transformer model and adaptive control algorithm are introduced to achieve accurate prediction and dynamic regulation of the working frequency of the reader in the RFID system. By comprehensively analyzing environmental factors, tag characteristics, reader characteristics and time series data, the present invention can effectively predict the optimal working frequency of the reader, and use methods such as PID controller and low-pass filter to ensure the stability and accuracy of frequency adjustment. This process not only improves the communication quality and stability of the RFID system, but also enhances the system's adaptability under different environmental conditions, ensuring long-term stable operation.

[0048] Those skilled in the art will appreciate that although some embodiments herein include certain features included in other embodiments but not other features, the combination of features from different embodiments is meant to be within the scope of the present invention and to form different embodiments.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling the frequency of an RFID system for managing power grid tools, characterized in that: The method comprises: Collect historical data of RFID system; Performing data fusion on the historical data of the RFID system to obtain fused data; Based on the fused data, respectively, an embedding vector and a position encoding vector of the fused data are obtained through an embedding layer and a position encoding vector of a Transformer model, and input data of the Transformer model is constructed based on the embedding vector and the position encoding vector; Training a Transformer model based on input data of the Transformer model; The trained Transformer model is used to predict the operating frequency of the RFID reader, and the prediction result is used as the target frequency to regulate and constrain the operating frequency of the RFID reader.

2. A method for controlling the frequency of a power grid tool management RFID system according to claim 1, characterized in that: The historical data of the RFID system includes: environmental factor data, tag characteristic data, reader characteristic data and time series characteristic data; The environmental factor data include: temperature ,humidity , metal material distance ; The tag characteristic data includes: tag type , aging degree ; The reader characteristic data includes: reader power , reader position ; The time series characteristic data includes: frequency data , timestamp ; The step of fusing the historical data of the RFID system to obtain fused data includes: Preprocessing the historical data of the RFID system; The discrete feature data in the preprocessed data is encoded through one-hot feature encoding; The discrete feature data after feature encoding is merged with the continuous feature data in the preprocessed data to form a unified feature vector; Fusing the time series feature data with the feature vector to obtain fused data; The discrete feature data includes: label type , reader position ; The continuous characteristic data includes: temperature ,humidity , metal material distance , aging degree , reader power , frequency data .

3. A method for controlling the frequency of a power grid tool management RFID system according to claim 2, characterized in that: The discrete feature data in the preprocessed data is encoded by one-hot, including: Select the discrete feature data that needs to be One-Hot encoded; According to the categories of the tag type and the categories of the reader position, a tag type binary vector and a reader position binary vector are created, wherein the length of the vector is equal to the number of feature categories, and each category corresponds to a position in the vector; the categories of the tag type include: passive tags, active tags, and semi-passive tags; the categories of the reader position include: entrance, exit, production line, and warehouse; The label type binary vector refers to a binary vector of length 3, expressed as: ； in, Refers to passive tags, refers to active tags, It refers to semi-passive tags; The reader position binary vector refers to a binary vector with a length of 4, expressed as: ； in, It refers to the entrance. It refers to the exit. It refers to the production line. It refers to the warehouse; The step of fusing the time series feature data with the feature vector to obtain fused data includes: The timestamps of the time series data are synchronized with the feature data in time, and the corresponding feature vector is found for each timestamp; the time series data and the feature vector are fused to obtain the fused data, which is expressed as: ； in, refers to the fused data. It refers to time The feature vector at time, It refers to time Frequency data at time, It refers to time The timestamp of the moment, It refers to the concatenation of multiple vectors or values.

4. A method for controlling the frequency of a power grid tool management RFID system according to claim 1, characterized in that: The method of obtaining an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and constructing input data of the Transformer model based on the embedding vector and the position encoding vector, comprises: Mapping the feature vector of the fused data to a high-dimensional space through the embedding layer to obtain an embedding vector; The position encoder adds the time position information to each time point to obtain a position encoding vector; The embedding vector and the position encoding vector are added together to obtain the input data of the Transformer model.

5. A method for controlling the frequency of a power grid tool management RFID system according to claim 4, characterized in that: The calculation formula of the position encoding vector is: ； ； in, is the position encoding vector, refers to the dimension index of the position encoding, refers to the dimension of the embedding vector.

6. A method for controlling the frequency of a power grid tool management RFID system according to claim 1, characterized in that: The Transformer model calculates the attention weights between input vectors through a multi-head self-attention mechanism at the encoder layer; performs nonlinear transformation on the output of the multi-head self-attention mechanism, adds residual connections after each sub-layer, and performs layer normalization, and finally constructs a deep feature representation by stacking multiple encoder layers; The Transformer model decoder output is mapped to the target prediction space, and the frequency value of the prediction result is output by the linear layer.

7. A method for controlling the frequency of a power grid tool management RFID system according to claim 6, characterized in that: The attention head calculation formula of the multi-head self-attention mechanism is: ； in, is the dot product of the query matrix and the key matrix, is the scaling factor, is the dimension of the key vector, is a matrix of values; The calculation formula of the nonlinear transformation is: ； in, is the input vector, and is the weight matrix, and is the bias term, It refers to the activation function.

8. A method for controlling the frequency of a power grid tool management RFID system according to claim 1, characterized in that: The training of the Transformer model based on the input data of the Transformer model includes: Use mean squared error as the loss function; Use Adam optimizer to update parameters; Perform forward propagation on the training set, calculate the loss, perform backpropagation and parameter update, iteratively train the model until the loss function converges, and test the model on the validation set to calculate the validation loss.

9. A method for controlling the frequency of a power grid tool management RFID system according to claim 1, characterized in that: The method of using the prediction result as the target frequency to regulate and constrain the working frequency of the RFID reader includes: Set the working frequency of the reader / writer according to the prediction results, use the prediction results as the target frequency, define the upper and lower limits of the reader / writer frequency adjustment, and ensure that the frequency adjustment does not exceed the working range of the device; Based on the discrete time system, a dynamic model of the reader's operating frequency is constructed, and the error between the target frequency and the actual frequency is calculated; The PID controller is used to compensate the frequency error, and the gain coefficient of the PID controller is adjusted in real time based on the changes in the system response; The control signal is smoothed by a low-pass filter; the filtered control signal is applied to the system to update the working frequency of the reader; Calculate the fluctuation range and error variance of the reader frequency, evaluate the frequency stability after adjustment, update the system model and controller parameters through adaptive algorithms based on the adjustment results, and dynamically adjust the key parameters of the system model and controller according to the feedback data during operation; Continuously optimize control algorithms through real-time monitoring and feedback mechanisms; The dynamic model expression is: ； in, Refers to the control input, which means that at time The adjustment amount of the reader frequency at all times, It refers to the actual operating frequency; The error calculation formula is: ； in, Refers to the error between the target frequency and the actual frequency, which is used to adjust the control input. is the target frequency, It refers to the actual operating frequency; The output of the PID controller is: ； in, is the proportional gain, is the integral gain, is the differential gain, is the cumulative sum of errors, It refers to the rate of change of error; The smoothing expression is: ； in, is the filtered control input, is the filter coefficient, is the control input, Refers to the filter control input at the previous moment; The working frequency of the update reader is expressed as: ； in, is the actual operating frequency, Refers to the filtered control input.

10. A frequency control system for power grid tool management RFID system, characterized in that: The system comprises: A data processing module is used to collect historical data of the RFID system, perform data fusion on the historical data of the RFID system to obtain fused data, obtain an embedding vector and a position encoding vector of the fused data through an embedding layer and a position encoder of a Transformer model based on the fused data, and construct input data of the Transformer model based on the embedding vector and the position encoding vector; A training module, used for training a Transformer model based on input data of the Transformer model; The control module is used to use the trained Transformer model to predict the operating frequency of the RFID reader, and use the prediction result as the target frequency to regulate and constrain the operating frequency of the RFID reader.

Images