A data analysis method for platform scales based on Internet of Things technology

Abstract

The present invention discloses a platform scale data analysis method based on the Internet of Things technology, which relates to the field of Internet of Things data processing technology, including collecting and preprocessing platform scale data, calculating fusion feature vectors based on the preprocessed platform scale data; predicting future fusion feature vectors, and calculating the platform scale health index based on the future fusion feature vectors; constructing a long short-term memory LSTM model to predict the remaining life of the platform scale; constructing a visualization interface to display the remaining life of the platform scale, and storing, collecting and analyzing the generated platform scale data. The wavelet coefficients and the normalized mutual information are weightedly fused to generate a fusion feature vector, which solves the problem of insufficient data processing and prediction accuracy in the prior art, improves the ability to recognize implicit patterns in platform scale data, combines the maximum Lyapunov exponent method and the nonlinear adjustment factor of the Logistic mapping to predict future fusion feature vectors, and improves the accuracy and prediction ability of platform scale data analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things data processing, and in particular to a platform scale data analysis method based on Internet of Things technology. Background Art

[0002] With the rapid development of Internet of Things technology, the demand for accurate data collection and analysis in various industries is growing, especially in the fields of industry and logistics. The popularization of Internet of Things applications has promoted the rapid development of smart devices. Platform scales, as common industrial weighing equipment, are widely used in material transportation, warehouse management, production line monitoring and other occasions. Their accuracy and stability have an important impact on production and logistics efficiency. Platform scale data usually includes load, temperature and other information. These data can provide key information for equipment health monitoring, fault warning and performance optimization. The processing of platform scale data involves multiple dimensions, including dynamic environmental changes and equipment usage status. Therefore, a more intelligent data analysis method is needed to achieve more accurate monitoring and prediction.

[0003] The existing technology still has certain limitations in the processing of scale data. The method of data feature extraction is relatively single, lacks in-depth exploration of time-varying characteristics and nonlinear relationships, and cannot fully reflect the dynamic characteristics of equipment operation. The existing methods lack in-depth correlation analysis of load data and environmental data, and ignore the potential connection between load changes and environmental changes, which makes it difficult to achieve accurate load prediction and equipment health status assessment. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a scale data analysis method based on Internet of Things technology, which solves the problem that the way of extracting data features is relatively single, lacks in-depth mining of time-varying features and nonlinear relationships, and cannot fully reflect the dynamic characteristics of equipment operation. The existing methods lack in-depth correlation analysis of load data and environmental data, and ignore the potential connection between load changes and environmental changes, which makes it difficult to achieve accurate load prediction and equipment health status assessment.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a scale data analysis method based on Internet of Things technology, which includes collecting scale data and preprocessing it, calculating a fused feature vector based on the preprocessed scale data; predicting future fused feature vectors, and calculating a scale health index based on the future fused feature vectors; constructing a long short-term memory (LSTM) model to predict the remaining life of the scale; constructing a visual interface to display the remaining life of the scale, and storing, collecting and analyzing the generated scale data.

[0008] As a preferred solution of the platform scale data analysis method based on the Internet of Things technology of the present invention, wherein: the collecting and preprocessing of platform scale data refers to using an Internet of Things sensor to collect and preprocess the platform scale data;

[0009] The IoT sensor includes a strain gauge and a temperature sensor;

[0010] The scale data includes load and temperature data;

[0011] The preprocessing includes using a time synchronization protocol to synchronize the load and temperature data, using a Kalman filter to perform denoising, using a box plot method to identify and delete abnormal data, using a multiple interpolation method to fill in missing data, and standardizing the scale data.

[0012] As a preferred solution of the scale data analysis method based on the Internet of Things technology of the present invention, wherein: the calculation of the fusion feature vector based on the pre-processed scale data refers to using the mother wavelet of the Daubechies wavelet as the wavelet function , where t is time;

[0013] Based on the pre-processed background scale data, it is defined as , ,in is the load data, For temperature data, discrete wavelet transform is used to perform wavelet decomposition to obtain wavelet coefficients , where j is the scale index of wavelet decomposition;

[0014] The pre-processed load and temperature data are discretized using the equidistant binning method, and the joint probability distribution is calculated using the frequency counting method. ,in and are the discretized values ​​of load and temperature data respectively;

[0015] Based on the joint probability distribution , use the marginalization method to calculate the marginal probability of the load and the marginal probability of the temperature data ;

[0016] Calculate the mutual information between load data and temperature data using mutual information analysis ;

[0017] Use Shannon entropy to calculate the entropy of the load separately and the entropy of the temperature data ;

[0018] Mutual Information Normalize it and get the normalized mutual information ;

[0019] According to the wavelet coefficients and normalized mutual information , using the weighted sum fusion method to calculate the fusion feature vector , the formula is:

[0020] ,

[0021] Where J is the number of decomposition levels.

[0022] As a preferred solution of the scale data analysis method based on the Internet of Things technology of the present invention, the predicted future fusion feature vector refers to using a dynamic time warping algorithm to calculate the fusion feature vector The alignment distance between each pair of feature components in , where o and s are the indices of the characteristic components;

[0023] Based on the alignment distance between each pair of feature components Generate an alignment distance matrix, align the feature components in time, obtain the aligned feature components, concatenate them in the order after time alignment, generate an aligned feature matrix, and record it as the comprehensive state space trajectory;

[0024] Randomly select rows from the comprehensive state space trajectory, record them as complex network nodes, use the Euclidean distance method to calculate the Euclidean distance between nodes, use the k-NN algorithm to connect the nodes, record them as complex network edges, use the inverse proportional weighting method to calculate the weight of the complex network edge, define the weight of the complex network edge as the element of the adjacency matrix, and construct the adjacency matrix P;

[0025] Based on the adjacency matrix, the node degree is calculated using weighted degree, and the transition probability between nodes is calculated using the random walk model to generate a transition probability matrix;

[0026] The control parameter r is calculated using the maximum Lyapunov exponent method, and the nonlinear adjustment factor is calculated using the Logistic mapping formula. ;

[0027] Based on nonlinear adjustment factor , use the Markov chain prediction formula to predict future time The fusion feature vector , the formula is:

[0028] ,

[0029] in is the u-th power of the transition probability matrix P, is the time step.

[0030] As a preferred solution of the scale data analysis method based on the Internet of Things technology of the present invention, wherein: the calculation of the scale health index based on the future fusion feature vector refers to the calculation of the scale health index based on the prediction of the future time. The fused feature vector is used to calculate the load deviation using the Euclidean distance method. , using the normalized load deviation scoring method to convert the load deviation Converted to load deviation score ;

[0031] Collect historical temperature data and preprocess them, and use statistical analysis to calculate the mean of the temperature data and set it as the indicator of the temperature data ;

[0032] Based on temperature data and indicators , using the percentage deviation method to calculate the environmental impact index ;

[0033] Scoring based on load deviation and Environmental Impact Index , using the geometric mean method to calculate the scale health index .

[0034] As a preferred solution of the platform scale data analysis method based on the Internet of Things technology described in the present invention, wherein: the construction of the long short-term memory LSTM model to predict the remaining life of the platform scale refers to collecting historical platform scale data and preprocessing it, using a weighted comprehensive evaluation method to calculate the historical platform scale health index, and generating a training set;

[0035] Build a long short-term memory (LSTM) model, including input layer, LSTM layer, and fully connected output layer;

[0036] Define the input layer as the scale health index;

[0037] Use the training set to train the LSTM model, and use the loss function and Adam optimizer to iteratively optimize the model parameters;

[0038] The scale health index Input into the trained long short-term memory (LSTM) model to predict the remaining life of the scale.

[0039] As a preferred solution of the scale data analysis method based on the Internet of Things technology of the present invention, wherein: the construction of a visual interface to display the remaining life of the scale refers to using the visualization tool Matplotlib to construct a visual interface, laying out a chart area in the middle of the page, displaying the remaining life of the scale in real time, adding an adjustment bar to the sidebar of the page, and displaying the health index of the scale;

[0040] Only users who have passed real-name verification are allowed to view it.

[0041] As a preferred solution of the scale data analysis method based on Internet of Things technology described in the present invention, wherein: the storage of the scale data collected and analyzed refers to storing the collected scale data and the remaining life of the scale generated by the analysis in a central database, and setting security access measures. The central database backs up the stored data in the cloud, and regularly performs integrity checks on the stored data and backup data. After the test is completed, an integrity test record is generated and stored synchronously in the central database.

[0042] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the scale data analysis method based on the Internet of Things technology as described in the first aspect of the present invention is implemented.

[0043] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the scale data analysis method based on Internet of Things technology as described in the first aspect of the present invention is implemented.

[0044] The beneficial effects of the present invention are: by collecting scale data and preprocessing it, a fusion feature vector is calculated based on the preprocessed scale data; future fusion feature vectors are predicted, and a scale health index is calculated based on the future fusion feature vectors; the accuracy and prediction ability of scale data analysis are improved, data processing capabilities are enhanced, and the ability to recognize implicit patterns in scale data is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0046] Figure 1 This is a flow chart of the scale data analysis method based on Internet of Things technology in Example 1.

[0047] Figure 2 This is a schematic diagram of fusion feature vector calculation of the scale data analysis method based on Internet of Things technology in Example 1. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0051] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a method for analyzing platform scale data based on Internet of Things technology, comprising the following steps:

[0052] S1, collect and preprocess the scale data, and calculate the fusion feature vector based on the preprocessed scale data;

[0053] Specifically, collecting and preprocessing the scale data refers to using an IoT sensor to collect and preprocess the scale data;

[0054] The IoT sensor includes a strain gauge and a temperature sensor;

[0055] The scale data includes load and temperature data;

[0056] The preprocessing includes using a time synchronization protocol to synchronize the load and temperature data, using a Kalman filter to perform denoising, using a box plot method to identify and delete abnormal data, using a multiple interpolation method to fill in missing data, and standardizing the scale data.

[0057] The Internet of Things (IoT) sensor is an intelligent sensing device integrated in the IoT system, which can collect and transmit data in the physical environment in real time. Through the time synchronization protocol, it ensures that the load data and temperature data are completely consistent in time. In the scale data, the sensor is often disturbed by the external environment and generates noise signals. The Kalman filter can accurately filter out these noises, retain the main features of the signal, and ensure that the load data and temperature data obtained in the subsequent analysis are more accurate. By cleaning the outliers in the data, the present invention can ensure the quality of the remaining data and provide more accurate input data for subsequent feature extraction and data modeling, thereby improving the effectiveness of the prediction model. By using the multiple interpolation method, these missing values ​​can be effectively filled while retaining the data structure and distribution characteristics, avoiding the impact of missing data on the analysis results. Through standardization processing, the scale load and temperature data are converted into dimensionless and unified standard data, eliminating the scale differences between the data, ensuring that the subsequent feature extraction and analysis are more stable and reliable.

[0058] Further, the fusion feature vector is calculated based on the preprocessed scale data using the mother wavelet of Daubechies wavelet as the wavelet function , the formula is:

[0059] ,

[0060] Where t is time;

[0061] Based on the pre-processed background scale data, it is defined as , ,in is the load data, For temperature data, discrete wavelet transform is used to perform wavelet decomposition to obtain wavelet coefficients , the formula is:

[0062] ,

[0063] Where J is the number of decomposition levels, j is the scale index of wavelet decomposition;

[0064] The pre-processed load and temperature data are discretized using the equidistant binning method, and the joint probability distribution is calculated using the frequency counting method. ,in and are the discretized values ​​of load and temperature data respectively;

[0065] Based on the joint probability distribution , use the marginalization method to calculate the marginal probability of the load and the marginal probability of the temperature data ;

[0066] Calculate the mutual information between load data and temperature data using mutual information analysis , the formula is:

[0067] ,

[0068] in and are the number of discretized intervals of load and temperature data, respectively, and i and l are the indices of all discretized values, respectively;

[0069] Use Shannon entropy to calculate the entropy of the load separately and the entropy of the temperature data ;

[0070] Mutual Information Normalize it and get the normalized mutual information , the formula is:

[0071] ,

[0072] According to the wavelet coefficients and normalized mutual information , using the weighted sum fusion method to calculate the fusion feature vector , the formula is:

[0073] .

[0074] Existing feature extraction methods usually rely on only a single feature, such as frequency domain features (Fourier transform) or time domain features (direct statistical features). Multi-scale features are extracted through wavelet transform, and further combined with normalized mutual information to measure the nonlinear relationship between features, which has obvious advantages in the comprehensiveness of time-frequency domain information. The multi-scale features provided by wavelet transform can reflect the local and global changes of load data, while mutual information analysis can quantify the dependency between load and environmental data. Existing feature fusion methods usually use simple linear fusion or average calculation, which fails to fully consider the importance of different features in the final analysis. By introducing wavelet coefficients as weights and dynamically adjusting the importance of features, the fused features are more in line with the operating characteristics and data characteristics of the actual equipment. The introduction of weights can highlight the features that have a greater impact on predicting the health status and life of the scale and weaken the role of irrelevant features, significantly improving the accuracy of the model. Through normalization processing, the fairness and comparability of the correlation between different features are ensured, making the fusion result more robust. Compared with some complex nonlinear feature fusion methods (such as high-dimensional features extracted by deep learning models), this method can complete feature fusion through simple weighted sum calculations, with higher computational efficiency, and is suitable for industrial applications with high real-time requirements. It solves the problems of single feature extraction, insufficient quantification of the relationship between features, and lack of flexibility of the fusion method in the existing technology. Compared with the existing technology, this method has higher feature representation ability, stronger robustness and higher computational efficiency;

[0075] By performing wavelet decomposition on the load and temperature data of the platform scale, the signal can be converted from the time domain to the frequency domain, thereby extracting data features at different frequencies. The wavelet coefficients can not only capture the low-frequency trend of the signal, but also effectively identify high-frequency fluctuations, especially when the load changes and ambient temperature fluctuates greatly. This decomposition method can highlight its inherent regularity. Data discretization processing through equidistant binning and frequency counting method can not only convert the load and temperature data into discrete forms, but also reveal the statistical correlation between the two through joint probability distribution. By calculating the mutual information between the load and temperature data, the correlation between them can be quantified, thereby identifying the degree of dependence of the two variables at the information level. Through the calculation of Shannon entropy, the complexity and uncertainty of each data source can be evaluated to help identify the most informative features. Finally, through weighted and fusion methods, the features of different data sources are integrated to obtain a comprehensive fusion feature vector, which improves the accuracy and reliability of platform scale data analysis.

[0076] S2, predicting the future fusion feature vector, and calculating the scale health index based on the future fusion feature vector;

[0077] Specifically, predicting the future fusion feature vector refers to using the dynamic time warping algorithm to calculate the fusion feature vector The alignment distance between each pair of feature components in , the formula is:

[0078] ,

[0079] in is the time alignment path, K is the time series length, o and s are the indexes of the feature components, k is the time alignment path index, and The characteristic components and At the point in time and The mapping value of

[0080] Based on the alignment distance between each pair of feature components Generate an alignment distance matrix, align the feature components in time, obtain the aligned feature components, concatenate them in the order after time alignment, generate an aligned feature matrix, and record it as the comprehensive state space trajectory;

[0081] Randomly select rows from the comprehensive state space trajectory, record them as complex network nodes, use the Euclidean distance method to calculate the Euclidean distance between nodes, use the k-NN algorithm to connect the nodes, record them as complex network edges, use the inverse proportional weighting method to calculate the weight of the complex network edge, define the weight of the complex network edge as the element of the adjacency matrix, and construct the adjacency matrix P;

[0082] Based on the adjacency matrix, the node degree is calculated using weighted degree, and the transition probability between nodes is calculated using the random walk model to generate a transition probability matrix;

[0083] The control parameter r is calculated using the maximum Lyapunov exponent method, and the nonlinear adjustment factor is calculated using the Logistic mapping formula. , the formula is:

[0084] ,

[0085] Based on nonlinear adjustment factor , use the Markov chain prediction formula to predict future time The fusion feature vector , the formula is:

[0086] ,

[0087] in is the u-th power of the transition probability matrix P, is the time step;

[0088] u is calculated by dividing the time step by the time interval between each state transition.

[0089] Prediction methods (such as prediction based on simple Markov models) usually assume that state transitions are linear and ignore the nonlinear change characteristics in actual data. Dynamically adjusting the prediction path through nonlinear adjustment factors can better adapt to the complex dynamic behavior of scale data, thereby significantly improving the accuracy of predictions. Unlike traditional single-step prediction methods, this method uses the Markov chain transfer matrix for multi-step state prediction, which can capture long-term state change trends. This multi-step prediction capability is particularly suitable for equipment health monitoring and life prediction scenarios, and can provide advance information for equipment maintenance and management. By combining nonlinear adjustment factors and Markov chains, it can not only predict The numerical change of the feature vector can also dynamically adjust the importance of the feature. This method of dynamic adjustment of fusion features is rare in the existing technology and is particularly suitable for time series prediction of multi-dimensional data. Compared with the method that only uses linear models or static models, this method can cope with the dynamic changes of nonlinear, multi-scale and multi-feature data and has stronger robustness. Based on Markov chain and simple matrix operations, it has higher computational efficiency than complex deep learning methods (such as RNN or LSTM). In industrial scenarios with large data volumes and the need for real-time prediction, it can significantly reduce the consumption of computing resources, improve the real-time performance of the system, and improve the accuracy of the remaining life prediction of platform scale equipment.

[0090] Using the dynamic time warping algorithm to align the fused feature vectors can effectively eliminate the impact of inconsistent time axes on data analysis, ensuring that different features can be compared and fused at the same time scale. By splicing the aligned feature components to generate a comprehensive state space trajectory, a multidimensional time series matrix can be formed, revealing the interaction and change trend between different features. Complex network modeling calculates the relationship between nodes through the Euclidean distance method, and uses the k-NN algorithm to connect nodes and construct an adjacency matrix, so that the model can better handle high-dimensional data, especially in the interaction analysis between multiple sensor data, which has strong application value. Combining the Logistic mapping formula for nonlinear adjustment further enhances the model's adaptability to complex dynamic behaviors, so that the prediction results can more accurately reflect the changing trend of the equipment's operating status. By calculating the Lyapunov exponent, the stability of the model in a dynamic environment can be accurately evaluated to avoid prediction errors caused by improper parameter selection. The Markov chain prediction formula can be used to predict future feature vectors based on the current state sequence and optimize the equipment health status prediction. The prediction method based on probability transfer can provide a reasonable estimate of the future state according to the evolution law of historical data, thereby providing decision support for equipment maintenance and load management.

[0091] Furthermore, the health index of the scale is calculated based on the future fusion feature vector. The fused feature vector is used to calculate the load deviation using the Euclidean distance method. , using the normalized load deviation scoring method to convert the load deviation Converted to load deviation score ;

[0092] Collect historical temperature data and preprocess them, and use statistical analysis to calculate the mean of the temperature data and set it as the indicator of the temperature data ;

[0093] Based on temperature data and indicators , using the percentage deviation method to calculate the environmental impact index , the formula is:

[0094] ,

[0095] Scoring based on load deviation and Environmental Impact Index , using the geometric mean method to calculate the scale health index , the formula is:

[0096] .

[0097] By using the Euclidean distance method to calculate load deviation, the difference between the actual load and the predicted load can be accurately evaluated. By using the statistical analysis method to calculate the mean of historical temperature data and using it as a benchmark indicator, the performance of the equipment under different environmental conditions can be clearly understood. Based on the normalized load deviation score and the environmental impact index, the geometric mean method is used to calculate the scale health index, which can comprehensively consider the impact of load deviation and environmental factors to obtain a more comprehensive health assessment result. The application of the geometric mean method makes the health index more robust and will not be affected by a single abnormal data. It can effectively reflect the overall health status of the scale and improve the reliability and accuracy of the scale.

[0098] S3, construct a long short-term memory (LSTM) model to predict the remaining life of the scale;

[0099] Specifically, constructing a long short-term memory (LSTM) model to predict the remaining life of a platform scale involves collecting historical platform scale data and preprocessing it, using a weighted comprehensive evaluation method to calculate the historical platform scale health index, and generating a training set;

[0100] Build a long short-term memory (LSTM) model, including input layer, LSTM layer, and fully connected output layer;

[0101] Define the input layer as the scale health index;

[0102] Use the training set to train the LSTM model, and use the loss function and Adam optimizer to iteratively optimize the model parameters;

[0103] The scale health index Input into the trained long short-term memory (LSTM) model to predict the remaining life of the scale.

[0104] Before training the LSTM model, by collecting historical scale data and preprocessing it, noise and outliers can be removed to ensure the quality of input data. The scale health index is calculated by the weighted comprehensive evaluation method, which can effectively integrate information from different sources (such as load data and temperature data) to obtain an indicator that comprehensively reflects the health status of the equipment. The LSTM model can learn the dynamic pattern of the scale health index changing over time, thereby effectively capturing the long-term dependency of the equipment status. The trained LSTM model can accurately predict the remaining life of the scale, thereby providing strong support for maintenance decisions. By inputting the historical health index, the model can give the remaining service life of the scale in the future, helping operators to perform maintenance in a timely manner, avoid equipment failures, and improve production efficiency.

[0105] S4, building a visual interface to display the remaining life of the scale, and storing the scale data collected and analyzed;

[0106] Specifically, building a visualization interface to display the remaining life of the scale refers to using the visualization tool Matplotlib to build a visualization interface, laying out a chart area in the middle of the page, displaying the remaining life of the scale in real time, and adding an adjustment bar in the sidebar of the page to display the health index of the scale;

[0107] Only users who have passed real-name verification are allowed to view it.

[0108] Using Matplotlib, real-time updated charts can be generated in the interface to display the remaining life of the scale. Through intuitive charts and data, users can more accurately determine whether equipment maintenance or replacement is needed, thereby improving equipment management efficiency and reducing the incidence of unexpected failures. By laying out the display area in the center of the page, users can see the remaining life data of the scale at a glance. Real-time display not only increases the convenience of user experience, but also can promptly reflect changes in the status of the equipment and help users respond quickly. The health index is a key factor affecting the remaining life of the scale. By displaying the health index in the sidebar, users can check the operating status of the equipment at any time. Through identity authentication, it ensures that only authorized users can view sensitive data such as the health status and remaining life of the equipment. This not only improves the security of the system and prevents information leakage, but also enables more accurate responsibility tracking during equipment management and maintenance, and improves the transparency and standardized management of equipment maintenance.

[0109] Furthermore, storing the scale data collected and analyzed means storing the collected scale data and the remaining life of the scale generated by the analysis in a central database, and setting up security access measures. The central database will back up the stored data to the cloud, and regularly perform integrity checks on the stored data and backup data. After the test is completed, an integrity test record will be generated and stored synchronously in the central database.

[0110] The collected scale data and remaining life analysis results are stored in the central database, providing a centralized data management platform. All data are stored in one place for easy access and processing at any time, avoiding data redundancy and management difficulties caused by distributed storage of multiple systems. Cloud backup can prevent data loss caused by hardware failure and ensure the long-term preservation and recoverability of equipment historical data. Through cloud storage, data can be restored in the event of catastrophic events (such as natural disasters, equipment failures, etc.) to ensure data security and integrity. Integrity testing ensures that data has not been illegally tampered with or damaged, thereby improving data credibility. By verifying the integrity of backup data, data loss or damage can be discovered in a timely manner, so that measures can be taken to repair or re-backup, avoiding potential risks caused by data loss. Secure data access control measures can enhance user trust in the system and improve system utilization and reliability.

[0111] This embodiment also provides a computer device, which is suitable for the case of a platform scale data analysis method based on the Internet of Things technology, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the platform scale data analysis method based on the Internet of Things technology proposed in the above embodiment.

[0112] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0113] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for analyzing scale data based on the Internet of Things technology proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0114] In summary, the present invention achieves the following results: collecting and preprocessing scale data, and calculating fused feature vectors based on the preprocessed scale data; predicting future fused feature vectors, and calculating the scale health index based on the future fused feature vectors; improving the accuracy and predictive ability of scale data analysis, enhancing data processing capabilities, and improving the ability to recognize implicit patterns in scale data.

[0115] 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A scale data analysis method based on Internet of Things technology, characterized in that: include, Collect and preprocess the scale data, and calculate the fusion feature vector based on the preprocessed scale data; Predict the future fusion feature vector, and calculate the scale health index based on the future fusion feature vector; Construct a long short-term memory (LSTM) model to predict the remaining life of a scale; Build a visual interface to display the remaining life of the scale, store, collect and analyze the scale data generated; The calculation of the fusion feature vector based on the preprocessed scale data refers to using the mother wavelet of the Daubechies wavelet as the wavelet function , the formula is: ， Where t is time; Based on the pre-processed background scale data, it is defined as , ,in is the load data, For temperature data, discrete wavelet transform is used to perform wavelet decomposition to obtain wavelet coefficients , the formula is: ， Where j is the scale index of wavelet decomposition; The pre-processed load and temperature data are discretized using the equidistant binning method, and the joint probability distribution is calculated using the frequency counting method. ,in and are the discretized values ​​of load and temperature data respectively; Based on the joint probability distribution , use the marginalization method to calculate the marginal probability of the load and the marginal probability of the temperature data ; Calculate the mutual information between load data and temperature data using mutual information analysis , the formula is: ， in and are the number of discretized intervals of load and temperature data, respectively, and i and l are the indices of all discretized values, respectively; Use Shannon entropy to calculate the entropy of the load separately and the entropy of the temperature data ; Mutual Information Normalize it and get the normalized mutual information , the formula is: ， According to the wavelet coefficient and normalized mutual information , using the weighted sum fusion method to calculate the fusion feature vector , the formula is: ， The predicting of the future fusion feature vector refers to using a dynamic time warping algorithm to calculate the fusion feature vector The alignment distance between each pair of feature components in , where o and s are the indices of the characteristic components; Based on the alignment distance between each pair of feature components Generate an alignment distance matrix, align the feature components in time, obtain the aligned feature components, concatenate them in the order after time alignment, generate an aligned feature matrix, and record it as the comprehensive state space trajectory; Randomly select rows from the comprehensive state space trajectory, record them as complex network nodes, use the Euclidean distance method to calculate the Euclidean distance between nodes, use the k-NN algorithm to connect the nodes, record them as complex network edges, use the inverse proportional weighting method to calculate the weight of the complex network edge, define the weight of the complex network edge as the element of the adjacency matrix, and construct the adjacency matrix P; Based on the adjacency matrix, the node degree is calculated using weighted degree, and the transition probability between nodes is calculated using the random walk model to generate a transition probability matrix; The control parameter r is calculated using the maximum Lyapunov exponent method, and the nonlinear adjustment factor is calculated using the Logistic mapping formula. , the formula is: ， Based on nonlinear adjustment factor , use the Markov chain prediction formula to predict future time The fusion feature vector , the formula is: ， in is the u-th power of the transition probability matrix P, is the time step; The calculation of the scale health index based on the future fusion feature vector refers to the calculation of the scale health index based on the prediction of the future time. The fused feature vector is used to calculate the load deviation using the Euclidean distance method. , using the normalized load deviation scoring method to convert the load deviation Converted to load deviation score ; Collect historical temperature data and preprocess them, and use statistical analysis to calculate the mean of the temperature data and set it as the indicator of the temperature data ; Based on temperature data and indicators , using the percentage deviation method to calculate the environmental impact index ; Scoring based on load deviation and Environmental Impact Index , using the geometric mean method to calculate the scale health index , the formula is: 。 2. The method for analyzing platform scale data based on Internet of Things technology according to claim 1, characterized in that: The collecting and preprocessing of platform scale data refers to using IoT sensors to collect and preprocess platform scale data; The IoT sensor includes a strain gauge and a temperature sensor; The scale data includes load and temperature data; The preprocessing includes using a time synchronization protocol to synchronize the load and temperature data, using a Kalman filter to perform denoising, using a box plot method to identify and delete abnormal data, using a multiple interpolation method to fill in missing data, and standardizing the scale data.

3. The method for analyzing platform scale data based on Internet of Things technology according to claim 1, characterized in that: The construction of the long short-term memory (LSTM) model to predict the remaining life of a platform scale refers to collecting historical platform scale data and preprocessing it, using a weighted comprehensive evaluation method to calculate the historical platform scale health index, and generating a training set; Build a long short-term memory (LSTM) model, including input layer, LSTM layer, and fully connected output layer; Define the input layer as the scale health index; Use the training set to train the LSTM model, and use the loss function and Adam optimizer to iteratively optimize the model parameters; The scale health index Input into the trained long short-term memory (LSTM) model to predict the remaining life of the scale.

4. The method for analyzing platform scale data based on Internet of Things technology according to claim 3, characterized in that: The said constructing a visualization interface to display the remaining life of the scale refers to using the visualization tool Matplotlib to construct a visualization interface, laying out a chart area in the middle of the page, displaying the remaining life of the scale in real time, and adding an adjustment bar in the sidebar of the page to display the health index of the scale; Only users who have passed real-name verification are allowed to view it.

5. The method for analyzing platform scale data based on Internet of Things technology according to any one of claims 1 and 3, characterized in that: The storage, collection and analysis of the scale data refers to storing the collected scale data and the remaining life of the scale generated by the analysis in a central database, and setting up security access measures. The central database will back up the stored data in the cloud, and regularly perform integrity checks on the stored data and backup data. After the test is completed, an integrity test record will be generated and stored synchronously in the central database.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the platform scale data analysis method based on Internet of Things technology described in any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the scale data analysis method based on Internet of Things technology described in any one of claims 1 to 5 are implemented.

Images