Electric tool data analysis and detection system and method based on multiple sensors

Abstract

The invention discloses an electric tool data analysis and detection system and method based on multiple sensors, and relates to the technical field of electric tools, the system comprises a data processing module, a data analysis module, a fault diagnosis module and a fault evaluation module; the data processing module is used for receiving original data from a plurality of sensors, performing preprocessing and data fusion on the original data and transmitting the data to the data analysis module; the data analysis module is used for receiving data of the data processing module, extracting characteristics capable of representing the running state of the electric tool from the data and transmitting the characteristics to the fault diagnosis module; the fault diagnosis module is used for performing fault diagnosis by adopting a method of combining a support vector machine and a convolutional neural network based on the features extracted by the data analysis module, outputting a diagnosis result and transmitting the diagnosis result to the fault evaluation module; and the fault evaluation module is used for carrying out quantitative evaluation on the diagnosed fault and setting graded early warning for reminding.

Description

Technical Field

[0001] The present invention relates to the technical field of electric tools, and in particular to a multi-sensor based electric tool data analysis and detection system and method. Background Art

[0002] In modern industrial and household application environments, power tools are key production and maintenance equipment, and their performance stability and reliability are directly related to production efficiency, product quality and safety of use. However, with the increase in the use time of power tools and the complexity and changeability of the working environment, such as high-intensity vibration, temperature changes, overload use and other factors, power tools often have various faults, such as motor overheating, bearing wear, transmission system failure, etc. These faults will not only affect the normal use of the tools, but may also cause safety hazards to the operators.

[0003] Traditional power tool fault detection methods mainly rely on manual inspections and regular maintenance, which have obvious limitations. First, manual inspections are difficult to fully cover all potential fault points and can easily miss subtle signs of faults. Second, although regular maintenance can detect some obvious faults, it is often not performed until the faults have developed to a certain extent, making it impossible to achieve early warning and timely elimination of faults. Finally, traditional methods lack data support and are difficult to accurately determine the type and severity of faults, resulting in low maintenance efficiency and high costs. Summary of the invention

[0004] The object of the present invention is to provide a multi-sensor based electric tool data analysis and detection system and method to solve the problems raised in the prior art.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a multi-sensor-based power tool data analysis and detection system, the system comprising a data processing module, a data analysis module, a fault diagnosis module, and a fault assessment module;

[0006] Furthermore, the data processing module is used to receive raw data from multiple sensors, process the raw data and transmit it to the data analysis module, and specifically includes a data preprocessing unit and a data fusion unit; the data preprocessing unit preprocesses the raw data of multiple sensors, including filtering, denoising, and calibration, and simultaneously performs time synchronization and spatial alignment on the data; the data fusion unit uses a Kalman filter algorithm to perform data fusion.

[0007] In the above technical solution, the raw data of multiple sensors are preprocessed to remove noise and interference in the raw data, achieve accurate calibration of the data and alignment of time and space, and use the Kalman filter algorithm to further fuse multiple sensor data to generate a more comprehensive and accurate data set, laying a solid foundation for subsequent data analysis. This process improves the availability and reliability of the data.

[0008] Further, the data analysis module receives data from the data processing module, extracts the characteristics of the operating state of the power tool and integrates them into a characteristic data set, including a time domain characteristic unit, a frequency domain characteristic unit, a time-frequency domain characteristic unit, a statistical characteristic unit, and an advanced characteristic unit;

[0009] The time domain feature unit is used to display the basic statistical characteristics of the power tool during operation and extract features directly related to the time series;

[0010] The frequency domain feature unit is used to identify the vibration mode and harmonic components in the operation of the electric tool, and extract the frequency domain features by performing Fourier transform on the data;

[0011] The time-frequency domain feature unit is used to capture the dynamic characteristics of the electric tool in different operating stages or working conditions, and extracts the frequency characteristics that change with time by combining time-domain and frequency-domain analysis and adopting short-time Fourier transform and wavelet transform methods;

[0012] The statistical feature unit is used to reveal nonlinear relationships, dynamic changes, periodicity, outliers or interdependence between data and multidimensional, nonlinear, time-varying or mutually coupled associations between elements within the data, and extracts features of high-order statistics, entropy, and correlation by comprehensively using high-order statistics, entropy measurement, and statistical methods of correlation analysis;

[0013] The advanced feature unit extracts features directly related to the fault, including temperature change trends, current fluctuations, etc., according to the specific application scenarios of the power tool.

[0014] In the above technical solution, multiple feature extraction units are integrated to realize multi-dimensional and multi-level analysis of the power tool operation data. Through the comprehensive use of time domain, frequency domain, time-frequency domain and statistical characteristics, the operation characteristics of the power tool can be revealed from different angles. It also combines the specific application scenarios of the power tool to extract targeted fault features, providing rich and accurate data support for fault diagnosis.

[0015] Further, the fault diagnosis module, based on the features extracted by the data analysis module, uses a method combining support vector machine and convolutional neural network to perform fault diagnosis, and uses a method combining support vector machine and convolutional neural network to perform power tool fault diagnosis, including a support vector machine unit, a convolutional neural network unit, and a comprehensive output unit;

[0016] First, the feature data set of the data analysis module is divided into a training set, a validation set, and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training;

[0017] The support vector machine unit can distinguish different types of data points by optimizing the decision boundary during model training, and can distinguish various fault states of the power tool from normal states. At the same time, a kernel function is introduced to handle nonlinear classification problems, and finally a preliminary prediction result including fault type classification, fault location, fault characteristics and corresponding fault probability is output;

[0018] The convolutional neural network unit, based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; after receiving the data at the input layer, it automatically learns and extracts deeper features, including local details of the fault signal and the evolution law of the fault in space and time through the alternating action of multiple convolutional layers and pooling layers; finally, the fully connected layer integrates the extracted features as the basis for classification and outputs the prediction results, including fault type classification, fault location, fault characteristics and corresponding fault probability;

[0019] The comprehensive output unit is used to summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; according to the performance indicators of the models in training, different weights are assigned to the prediction results of the two models, where w z is the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

[0020] In the above technical solution, by combining the advantages of support vector machines and convolutional neural networks, efficient and accurate diagnosis of power tool faults is achieved. The support vector machine performs preliminary diagnosis with its powerful classification ability, while the convolutional neural network further explores the spatial and temporal correlation of the data and extracts deep-level features; finally, through the probability weighted average method of the comprehensive output unit, the fusion of the two prediction results greatly improves the accuracy and reliability of the diagnosis.

[0021] Furthermore, the fault assessment module includes a fault assessment unit and a fault warning unit, which quantitatively assesses the diagnosed faults and sets graded warnings for reminders;

[0022] The fault assessment unit performs quantitative assessment on the diagnostic results output by the fault diagnosis module, including vibration amplitude assessment, temperature anomaly assessment, current fluctuation assessment and comprehensive assessment, to further determine the fault risk level of the electric tool;

[0023] The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula:

[0024] VI=A v -A vt / A vmax -A vt ;

[0025] Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safety upper limit of the vibration amplitude. The value range of VI is [0, +∞), where VI = 0 means the vibration is within the normal range, and the larger the VI value, the more severe the vibration;

[0026] The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula:

[0027] TI=|TT p | / T max -T min ;

[0028] Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min They represent the highest and lowest temperatures allowed under normal operating conditions, respectively. The value range of TI is [0,1], where TI = 0 means the temperature is at the optimal state, and TI close to 1 means the temperature is close to the limit value;

[0029] The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula:

[0030]

[0031] in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point;

[0032] The standard deviation σ of the current is calculated according to the formula:

[0033]

[0034] Calculate the current fluctuation rate IC according to the formula:

[0035]

[0036] The smaller the IC value, the more stable the current; the larger the IC value, the greater the current fluctuation;

[0037] The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula:

[0038] F=w VI *VI+*w TI *TI+w IC *IC;

[0039] where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate;

[0040] The fault warning unit divides the fault risks of the power tool into different levels according to the evaluation results of the fault evaluation unit and the output of the fault analysis module, and promptly reminds through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range, and records the information of each warning including fault type, fault characteristics, fault probability, and fault location.

[0041] In the above technical solution, the fault assessment module is an in-depth application and immediate response to the diagnostic results. By quantitatively assessing the fault risk and combining it with a comprehensive assessment algorithm, a comprehensive assessment of the power tool status is achieved. At the same time, the fault risk level is divided according to the assessment results, and the user is reminded in a timely manner through an intuitive method of sound and light alarms and display prompts, which effectively avoids the impact of potential faults on production activities and ensures the safe and stable operation of the power tool. It not only improves the timeliness of fault handling, but also enhances the intelligence level of the system.

[0042] A multi-sensor based electric tool data analysis and detection method, the method comprising the following steps:

[0043] Step S100: collecting raw data from multiple sensors, preprocessing the raw data, including filtering, denoising, and calibration, while performing time synchronization and spatial alignment on the data, using the Kalman filter algorithm for data fusion, and finally generating a fused data set and transmitting it to the data analysis module;

[0044] Step S200: receiving a data set from a data processing module, extracting features of the operating state of the power tool therefrom and integrating them into a feature data set, wherein the features include: extracting features directly related to the time series; extracting features in the frequency domain by performing Fourier transform on the data; combining time domain and frequency domain analysis, using short-time Fourier transform and wavelet transform methods to extract frequency features that change over time; extracting features of high-order statistics, entropy, and correlation by comprehensively using high-order statistics, entropy measurement, and statistical methods of correlation analysis; extracting features directly related to faults according to specific application scenarios of the power tool;

[0045] Step S300 uses a combination of support vector machine and convolutional neural network to perform power tool fault diagnosis. The specific steps are as follows:

[0046] Step S301: firstly, the feature data set of the data analysis module is divided into a training set, a validation set and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training;

[0047] Step S302: In the support vector machine model training, the decision boundary is optimized to distinguish data points of different categories, so that various fault states of the power tool can be distinguished from the normal state. At the same time, the kernel function is introduced to deal with the nonlinear classification problem, and finally the preliminary prediction results including fault type classification, fault location, fault characteristics and corresponding fault probability are output;

[0048] Step S303: Based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; after receiving the data at the input layer, it automatically learns and extracts deeper features through the alternating effects of multiple convolutional layers and pooling layers, including local details of the fault signal and the evolution law of the fault in space and time; finally, the fully connected layer integrates the extracted features as the basis for classification and outputs preliminary prediction results, including fault type classification, fault location, fault characteristics and corresponding fault probability;

[0049] Step S304: Summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; assign different weights to the prediction results of the two models according to the performance indicators shown in the training, where w zis the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set

[0050] |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

[0051] Step S400 performs quantitative evaluation on the diagnosed fault and sets graded warnings for reminders. The specific steps are as follows:

[0052] Step S401: performing quantitative evaluation on the diagnostic results output by the fault diagnosis module, including vibration amplitude evaluation, temperature anomaly evaluation, current fluctuation evaluation and comprehensive evaluation, to further determine the fault risk level of the power tool;

[0053] The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula:

[0054] VI=A v -A vt / A vmax -A vt ;

[0055] Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safe upper limit of vibration amplitude;

[0056] The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula:

[0057] TI=|TT p | / T max -T min ;

[0058] Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min Respectively represent the maximum and minimum temperatures allowed under normal operating conditions;

[0059] The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula:

[0060]

[0061] in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point;

[0062] The standard deviation σ of the current is calculated according to the formula:

[0063]

[0064] Calculate the current fluctuation rate IC according to the formula:

[0065] IC = σ / I;

[0066] The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula:

[0067] F=w VI *VI+*w TI *TI+w IC *IC;

[0068] where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate;

[0069] Step S402: Based on the evaluation result of the fault evaluation unit and the output of the fault analysis module, the fault risk of the power tool is divided into different levels, and timely reminders are given through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range. At the same time, the information of each warning is recorded, including the fault type, fault characteristics, fault probability, and fault location.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] By integrating multiple sensors, the system can obtain comprehensive data on the operating status of the power tool, including information in multiple dimensions such as vibration, temperature, and current, which enables the system to more accurately identify possible faults that may occur during the operation of the power tool;

[0072] The present invention adopts a combination of support vector machine and convolutional neural network to perform fault diagnosis, which further improves the accuracy of fault detection. These two algorithms can automatically learn and extract deep features in the data, effectively distinguish between normal state and fault state, and the system can receive and analyze sensor data in real time. Once an abnormality is found, an early warning is immediately issued, thereby improving the timeliness of fault detection;

[0073] Through automated fault diagnosis and evaluation, the present invention can accurately and in real time identify various types of faults occurring during the operation of the power tool, and determine the risk level of the fault through quantitative evaluation. This can not only improve the efficiency and accuracy of fault identification, but also enable maintenance personnel to quickly locate the root cause of the problem and formulate scientific and reasonable maintenance strategies based on the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A system module diagram of a multi-sensor-based electric tool data analysis and detection system of the present invention;

[0075] Figure 2 The present invention is a method flow chart of a multi-sensor based electric tool data analysis and detection method. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution.

[0078] A multi-sensor based power tool data analysis and detection system, characterized in that: the system comprises a data processing module, a data analysis module, a fault diagnosis module, and a fault assessment module;

[0079] The data processing module is used to receive raw data from multiple sensors, process the raw data and transmit it to the data analysis module, and specifically includes a data preprocessing unit and a data fusion unit; the data preprocessing unit preprocesses the raw data of multiple sensors, including filtering, denoising, and calibration, and performs time synchronization and spatial alignment on the data; the data fusion unit uses the Kalman filter algorithm to perform data fusion, and first defines the state equation and observation equation of the system:

[0080] Equation of state: x k =Ax k-1 +Bu k-1 +w k-1 ;

[0081] where x k is the state vector at time k, x k-1 is the state vector at the previous moment, A is the state transfer matrix, B is the control input matrix, u k-1is the control input vector, w k-1 is the process noise;

[0082] Observation equation: z k =Hx k +u k ;

[0083] Among them, z k is the observation vector at time k, H is the observation matrix, u k is the observation noise;

[0084] Then the state vector and error covariance matrix are initialized. Then, in each time step, the algorithm receives observation data from different sensors. The algorithm uses the state estimate of the previous moment and the observation data of the current moment. These observation data contain information about the operating status of the power tool. Through the two steps of prediction and update, the state estimate is iteratively optimized. In the prediction stage, in each time step, the state estimate and covariance matrix of the previous moment are used to predict the state and its covariance matrix of the current moment. In the update stage, when new observation data is received, the Kalman gain is used to correct the prediction results, and the state estimate and covariance matrix are updated. Through repeated iteration of this process, the Kalman filter algorithm can effectively fuse multiple sensor data, and finally generate a fused data set and transmit it to the data analysis module.

[0085] In the above technical solution, the raw data of multiple sensors are preprocessed to remove noise and interference in the raw data, achieve accurate calibration of the data and alignment of time and space, and use the Kalman filter algorithm to further fuse multiple sensor data to generate a more comprehensive and accurate data set, laying a solid foundation for subsequent data analysis. This process improves the availability and reliability of the data.

[0086] The data analysis module receives data from the data processing module, extracts the characteristics of the operating state of the power tool and integrates them into a characteristic data set, including a time domain characteristic unit, a frequency domain characteristic unit, a time-frequency domain characteristic unit, a statistical characteristic unit, and an advanced characteristic unit;

[0087] The time domain feature unit is used to display the basic statistical characteristics of the power tool during operation and extract features directly related to the time series;

[0088] The frequency domain feature unit is used to identify the vibration mode and harmonic components in the operation of the electric tool, and extract the frequency domain features by performing Fourier transform on the data;

[0089] The time-frequency domain feature unit is used to capture the dynamic characteristics of the electric tool in different operating stages or working conditions, and extracts the frequency characteristics that change with time by combining time-domain and frequency-domain analysis and adopting short-time Fourier transform and wavelet transform methods;

[0090] The statistical feature unit is used to reveal nonlinear relationships, dynamic changes, periodicity, outliers or interdependence between data and multidimensional, nonlinear, time-varying or mutually coupled associations between elements within the data, and extracts features of high-order statistics, entropy, and correlation by comprehensively using high-order statistics, entropy measurement, and statistical methods of correlation analysis;

[0091] The advanced feature unit extracts features directly related to the fault, including temperature change trends, current fluctuations, etc., according to the specific application scenarios of the power tool.

[0092] In the above technical solution, multiple feature extraction units are integrated to realize multi-dimensional and multi-level analysis of the power tool operation data. Through the comprehensive use of time domain, frequency domain, time-frequency domain and statistical characteristics, the operation characteristics of the power tool can be revealed from different angles. It also combines the specific application scenarios of the power tool to extract targeted fault features, providing rich and accurate data support for fault diagnosis.

[0093] The fault diagnosis module uses a method combining support vector machine and convolutional neural network to perform fault diagnosis based on the features extracted by the data analysis module, and uses a method combining support vector machine and convolutional neural network to perform fault diagnosis on electric tools, including a support vector machine unit, a convolutional neural network unit, and a comprehensive output unit;

[0094] First, the feature data set of the data analysis module is divided into a training set, a validation set, and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training;

[0095] The support vector machine unit can distinguish different types of data points by optimizing the decision boundary during model training, and can distinguish various fault states of the power tool from normal states. At the same time, a kernel function is introduced to handle nonlinear classification problems, and finally a preliminary prediction result including fault type classification, fault location, fault characteristics and corresponding fault probability is output;

[0096] The convolutional neural network unit, based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; after receiving the data at the input layer, it automatically learns and extracts deeper features, including local details of the fault signal and the evolution law of the fault in space and time through the alternating action of multiple convolutional layers and pooling layers; finally, the fully connected layer integrates the extracted features as the basis for classification and outputs the prediction results, including fault type classification, fault location, fault characteristics and corresponding fault probability;

[0097] The comprehensive output unit is used to summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; according to the performance indicators of the models in training, different weights are assigned to the prediction results of the two models, where w z is the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

[0098] In the above technical solution, by combining the advantages of support vector machines and convolutional neural networks, efficient and accurate diagnosis of power tool faults is achieved. The support vector machine performs preliminary diagnosis with its powerful classification ability, while the convolutional neural network further explores the spatial and temporal correlation of the data and extracts deep-level features; finally, through the probability weighted average method of the comprehensive output unit, the fusion of the two prediction results greatly improves the accuracy and reliability of the diagnosis.

[0099] The fault assessment module includes a fault assessment unit and a fault warning unit, which performs quantitative assessment on the diagnosed faults and sets graded warnings for reminders;

[0100] The fault assessment unit performs quantitative assessment on the diagnostic results output by the fault diagnosis module, including vibration amplitude assessment, temperature anomaly assessment, current fluctuation assessment and comprehensive assessment, to further determine the fault risk level of the electric tool;

[0101] The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula:

[0102] VI=A v -A vt / A vmax -A vt ;

[0103] Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safety upper limit of the vibration amplitude. The value range of VI is [0, +∞), where VI = 0 means the vibration is within the normal range, and the larger the VI value, the more severe the vibration;

[0104] The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula:

[0105] TI=|TT p | / T max -T min ;

[0106] Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min They represent the highest and lowest temperatures allowed under normal operating conditions, respectively. The value range of TI is [0,1], where TI = 0 means the temperature is at the optimal state, and TI close to 1 means the temperature is close to the limit value;

[0107] The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula:

[0108]

[0109] in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point;

[0110] The standard deviation σ of the current is calculated according to the formula:

[0111]

[0112] Calculate the current fluctuation rate IC according to the formula:

[0113]

[0114] The smaller the IC value, the more stable the current; the larger the IC value, the greater the current fluctuation;

[0115] The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula:

[0116] F=w VI *VI+*w TI *TI+w IC *IC;

[0117] where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate;

[0118] The fault warning unit divides the fault risks of the power tool into different levels according to the evaluation results of the fault evaluation unit and the output of the fault analysis module, and promptly reminds through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range, and records the information of each warning including fault type, fault characteristics, fault probability, and fault location.

[0119] In the above technical solution, the fault assessment module is an in-depth application and immediate response to the diagnostic results. By quantitatively assessing the fault risk and combining it with a comprehensive assessment algorithm, a comprehensive assessment of the power tool status is achieved. At the same time, the fault risk level is divided according to the assessment results, and the user is reminded in a timely manner through an intuitive method of sound and light alarms and display prompts, which effectively avoids the impact of potential faults on production activities and ensures the safe and stable operation of the power tool. It not only improves the timeliness of fault handling, but also enhances the intelligence level of the system.

[0120] A multi-sensor based electric tool data analysis and detection method, characterized in that the method comprises the following steps:

[0121] Step S100: collecting raw data from multiple sensors, preprocessing the raw data, including filtering, denoising, and calibration, while performing time synchronization and spatial alignment on the data, using the Kalman filter algorithm for data fusion, and finally generating a fused data set and transmitting it to the data analysis module;

[0122] Step S200: used to receive a data set from a data processing module, extract features of the operating status of the power tool therefrom and integrate them into a feature data set, wherein the features include: extracting features directly related to the time series; extracting features in the frequency domain by performing Fourier transform on the data; combining time domain and frequency domain analysis, using short-time Fourier transform and wavelet transform methods to extract frequency features that change over time; extracting features of high-order statistics, entropy, and correlation by comprehensively using high-order statistics, entropy measurements, and statistical methods of correlation analysis; and extracting features directly related to faults based on specific application scenarios of the power tool.

[0123] Step S300 uses a combination of support vector machine and convolutional neural network to perform power tool fault diagnosis. The specific steps are as follows:

[0124] Step S301: firstly, the feature data set of the data analysis module is divided into a training set, a validation set and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training;

[0125] Step S302: In the support vector machine model training, the decision boundary is optimized to distinguish data points of different categories, so that various fault states of the power tool can be distinguished from the normal state. At the same time, the kernel function is introduced to deal with the nonlinear classification problem, and finally the preliminary prediction results including fault type classification, fault location, fault characteristics and corresponding fault probability are output;

[0126] Step S303: Based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; after receiving the data at the input layer, it automatically learns and extracts deeper features through the alternating effects of multiple convolutional layers and pooling layers, including local details of the fault signal and the evolution law of the fault in space and time; finally, the fully connected layer integrates the extracted features as the basis for classification and outputs preliminary prediction results, including fault type classification, fault location, fault characteristics and corresponding fault probability;

[0127] Step S304: Summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; assign different weights to the prediction results of the two models according to the performance indicators shown in the training, where w z is the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set

[0128] |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

[0129] Step S400 performs quantitative evaluation on the diagnosed fault and sets graded warnings for reminders. The specific steps are as follows:

[0130] Step S401: performing quantitative evaluation on the diagnostic results output by the fault diagnosis module, including vibration amplitude evaluation, temperature anomaly evaluation, current fluctuation evaluation and comprehensive evaluation, to further determine the fault risk level of the power tool;

[0131] The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula:

[0132] VI=A v -A vt / A vmax -A vt ;

[0133] Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safe upper limit of vibration amplitude;

[0134] The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula:

[0135] TI=|TT p | / T max -T min ;

[0136] Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min Respectively represent the maximum and minimum temperatures allowed under normal operating conditions;

[0137] The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula:

[0138]

[0139] in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point;

[0140] The standard deviation σ of the current is calculated according to the formula:

[0141]

[0142] Calculate the current fluctuation rate IC according to the formula:

[0143] IC = σ / I;

[0144] The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula:

[0145] F=w VI *VI+*w TI *TI+w IC *IC;

[0146] where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate;

[0147] Step S402: Based on the evaluation result of the fault evaluation unit and the output of the fault analysis module, the fault risk of the power tool is divided into different levels, and timely reminders are given through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range. At the same time, the information of each warning is recorded, including the fault type, fault characteristics, fault probability, and fault location.

[0148] In an embodiment of the present invention, taking an electrician's screwdriver as an example, raw data from an acceleration sensor, a temperature sensor, and a current sensor are first received, and then a filtering operation is performed on the raw data to remove high-frequency noise, a denoising operation is performed to remove random noise, and a calibration operation is performed to adjust the sensitivity according to the sensor specifications; denoising processing is performed on the data of the temperature sensor and the current sensor to ensure the accuracy of the data; and time synchronization and spatial alignment are performed on all sensor data to ensure data consistency.

[0149] Then, data fusion is performed to define the state equation and observation equation. Assuming that the state of the power tool includes vibration amplitude, temperature and current, which change over time, the current state is first predicted, including the predicted values ​​of vibration amplitude, temperature and current. Then, the Kalman gain is calculated, and the predicted state is adjusted according to the difference between the sensor observation value and the predicted value. The covariance of the state estimate is updated, and data fusion is completed. Finally, a fused data set is generated, including the estimated values ​​of vibration amplitude, temperature and current.

[0150] Then, feature extraction is performed on the fused data set. The time domain feature extracts the mean, standard deviation, etc. of the acceleration data. The frequency domain feature performs Fourier transform on the acceleration data to extract the main frequency components. The time-frequency domain feature uses short-time Fourier transform to capture the frequency changes in different time periods. The statistical feature calculates the high-order statistics and entropy of the current. The advanced feature extracts features directly related to motor wear and bearing failure for electric screwdrivers. Finally, a feature data set is formed and input into the fault diagnosis module.

[0151] Then, fault diagnosis was started. The feature data set was divided into training set, validation set and test set in a ratio of 6:2:2. The support vector machine model and convolutional neural network model were trained respectively. The support vector machine unit output the preliminary prediction results. Then, the convolutional neural network unit further analyzed the support vector machine and output a more detailed fault location and feature description. Finally, the probability weighted average method was used to combine the prediction results of the two models, and the final output was: "Bearing fault, located on the left bearing, vibration exceeded the standard, temperature was abnormal, and the probability of failure was 48%";

[0152] Based on the final diagnosis result of the fault diagnosis module, the fault assessment module further uses the data obtained from the sensor and processed by the data processing module to perform a quantitative fault assessment to determine the fault risk level of the power tool:

[0153] Vibration amplitude evaluation: Av = 1.5g, Avt = 1g, Avmax = 2g, then VI = (1.5-1) / (2-1) = 0.5;

[0154] Temperature anomaly assessment: T = 65°C, Tp = 50°C, Tmax = 70°C, Tmin = 40°C, then TI = |65-50| / (70-40) = 0.5;

[0155] Current fluctuation assessment: Set the number of sampling points N = 100 and calculate σ=0.1A, then IC=0.1 / 1.0=0.1;

[0156] Comprehensive evaluation: assign weight w VI =0.4, w TI =0.4, w IC =0.2, calculate F=0.4×0.5+0.4×0.5+0.2×0.1=0.42;

[0157] For the comprehensive evaluation result F, the following classification is set to quantify the failure risk:

[0158] Normal: F<0.2, indicating that the power tool operates within the normal range and there is no obvious sign of failure;

[0159] Low-level fault risk: 0.2≤F<0.4, indicating that the power tool has a slight abnormality, but it does not affect normal use. Regular inspection is recommended;

[0160] Medium failure risk: 0.4≤F<0.6, indicating that the power tool has a medium failure risk and needs close attention and preparation for maintenance or repair;

[0161] Advanced failure risk: 0.6≤F<0.8, indicating that the power tool has a high risk of failure, which may affect the safety of use. It is recommended to stop the power tool immediately for inspection and repair;

[0162] Emergency failure risk: F ≥ 0.8, indicating that the power tool has a serious failure risk, which may immediately cause damage or safety accidents, and must be shut down immediately and repaired urgently;

[0163] The evaluation result F=0.42 in this embodiment belongs to the medium fault risk range. The system will output and remind the user: "The left bearing of the electric screwdriver has a medium fault risk, and the failure probability is 48%. Please pay close attention and prepare for maintenance or repair!" At the same time, the system will also record the detailed information of this warning for subsequent analysis and processing.

[0164] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A multi-sensor based power tool data analysis and detection system, characterized in that: The system includes a data processing module, a data analysis module, a fault diagnosis module, and a fault assessment module; The data processing module is used to receive raw data from multiple sensors, process the raw data and transmit the raw data to the data analysis module; The data analysis module receives the data from the data processing module, extracts the features that can represent the operating status of the electric tool and transmits them to the fault diagnosis module; The fault diagnosis module uses a method combining support vector machine and convolutional neural network to perform fault diagnosis based on the features extracted by the data analysis module, and outputs the diagnosis results to the fault assessment module; The fault assessment module performs quantitative assessment on the diagnosed faults and sets graded warnings for reminders.

2. The multi-sensor based power tool data analysis and detection system according to claim 1, characterized in that: The data processing module includes a data preprocessing unit and a data fusion unit; The data preprocessing unit preprocesses the raw data of multiple sensors, including filtering, denoising, and calibration, and performs time synchronization and spatial alignment on the data; The data fusion unit uses a Kalman filter algorithm to perform data fusion, and finally generates a fused data set and transmits it to the data analysis module.

3. The multi-sensor based power tool data analysis and detection system according to claim 1, characterized in that: The data analysis module is used to receive the data set of the data processing module, extract the characteristics of the operating state of the power tool and integrate them into a characteristic data set, including a time domain characteristic unit, a frequency domain characteristic unit, a time-frequency domain characteristic unit, a statistical characteristic unit, and an advanced characteristic unit; The time domain feature unit is used to display the basic statistical characteristics of the power tool during operation and extract features directly related to the time series; The frequency domain feature unit is used to identify the vibration mode and harmonic components in the operation of the electric tool, and extract the frequency domain features by performing Fourier transform on the data; The time-frequency domain feature unit is used to capture the dynamic characteristics of the electric tool in different operating stages or working conditions, and extracts the frequency characteristics that change with time by combining time-domain and frequency-domain analysis and adopting short-time Fourier transform and wavelet transform methods; The statistical feature unit extracts the features of high-order statistics, entropy and correlation by comprehensively using the statistical method of high-order statistics, entropy measurement and correlation analysis; The advanced feature unit extracts features directly related to the fault according to the specific application scenario of the electric tool.

4. The multi-sensor based power tool data analysis and detection system according to claim 1, characterized in that: The fault diagnosis module uses a combination of support vector machine and convolutional neural network to perform power tool fault diagnosis, including a support vector machine unit, a convolutional neural network unit, and a comprehensive output unit; First, the feature data set of the data analysis module is divided into a training set, a validation set, and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training; The support vector machine unit can distinguish different types of data points by optimizing the decision boundary during model training, and can distinguish various fault states of the power tool from normal states. At the same time, a kernel function is introduced to handle nonlinear classification problems, and finally a preliminary prediction result including fault type classification, fault location, fault characteristics and corresponding fault probability is output; The convolutional neural network unit, based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; after receiving the data at the input layer, through the alternating action of multiple convolutional layers and pooling layers, it automatically learns and extracts deeper features, including local details of the fault signal and the evolution law of the fault in space and time; finally, the fully connected layer integrates the extracted features as the basis for classification, and outputs the prediction results including the fault type classification, fault location, fault characteristics and the corresponding fault probability; The comprehensive output unit is used to summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; according to the performance indicators of the models in training, different weights are assigned to the prediction results of the two models, where w z is the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

5. The multi-sensor based power tool data analysis and detection system according to claim 1, characterized in that: The fault assessment module includes a fault assessment unit and a fault warning unit; the fault assessment unit performs quantitative assessment on the diagnostic results output by the fault diagnosis module, including vibration amplitude assessment, temperature anomaly assessment, current fluctuation assessment and comprehensive assessment, to further determine the fault risk level of the electric tool; The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula: VI=A v -THE vt / THE vmax -THE vt ; Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safe upper limit of vibration amplitude; The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula: TI=|TT p | / T max -T min ; Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min Respectively represent the maximum and minimum temperatures allowed under normal operating conditions; The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula: in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point; The standard deviation σ of the current is calculated according to the formula: Calculate the current fluctuation rate IC according to the formula: The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula: F=w VI *VI+*w TI *TI+w IC *IC; where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate; The fault warning unit divides the fault risks of the power tool into different levels according to the evaluation results of the fault evaluation unit and the output of the fault analysis module, and promptly reminds through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range, and records the information of each warning including fault type, fault characteristics, fault probability, and fault location.

6. A multi-sensor based power tool data analysis and detection method, characterized in that: The method comprises the following steps: Step S100: collecting raw data from multiple sensors, processing the raw data and transmitting it to a data analysis module; Step S200: receiving data from the data processing module, extracting features that can represent the operating status of the power tool and transmitting them to the fault diagnosis module; Step S300: Based on the features extracted by the data analysis module, a method combining support vector machine and convolutional neural network is used to perform fault diagnosis, and the output diagnosis result is transmitted to the fault assessment module; Step S400: quantify and evaluate the diagnosed faults and set graded warnings for reminders.

7. The multi-sensor based power tool data analysis and detection method according to claim 6, characterized in that: The step S100 performs preprocessing and data fusion on the collected sensor data; Step S101: preprocessing the raw data of multiple sensors, including filtering, denoising, and calibration, and performing time synchronization and spatial alignment on the data; Step S102: using the Kalman filter algorithm to perform data fusion, and finally generating a fused data set and transmitting it to the data analysis module; The step S200 is used to receive a data set from a data processing module, extract the characteristics of the operating state of the power tool and integrate them into a characteristic data set, wherein the characteristics include: extracting characteristics directly related to the time series; extracting frequency domain characteristics by performing Fourier transform on the data; Combining time domain and frequency domain analysis, short-time Fourier transform and wavelet transform methods are used to extract frequency characteristics that change with time. The characteristics of high-order statistics, entropy measurement, and correlation analysis are extracted by comprehensively using statistical methods. According to the specific application scenarios of power tools, the characteristics directly related to the fault are extracted.

8. The multi-sensor based power tool data analysis and detection method according to claim 6, characterized in that: The step S300 uses a combination of support vector machine and convolutional neural network to perform power tool fault diagnosis, and the specific steps are as follows: Step S301: firstly, the feature data set of the data analysis module is divided into a training set, a validation set and a test set according to the system setting ratio, and the training set is input into the support vector machine model and the convolutional neural network model for training; Step S302: In the support vector machine model training, the decision boundary is optimized to distinguish data points of different categories, so that various fault states of the power tool can be distinguished from the normal state. At the same time, the kernel function is introduced to deal with the nonlinear classification problem, and finally the preliminary prediction results including fault type classification, fault location, fault characteristics and corresponding fault probability are output; Step S303: Based on the initial judgment of the support vector machine, the convolutional neural network model further analyzes the spatial and temporal correlation of the data; After receiving data at the input layer, multiple convolutional layers and pooling layers are used alternately to automatically learn and extract deeper features, including local details of the fault signal and the evolution of the fault in space and time. Finally, the fully connected layer integrates the extracted features as the basis for classification, and outputs prediction results including fault type classification, fault location, fault characteristics and corresponding fault probability. Step S304: Summarize the prediction results of the support vector machine and the convolutional neural network and use the probability weighted average method to make a comprehensive decision; assign different weights to the prediction results of the two models according to the performance indicators shown in the training, where w z is the weight of the assigned support vector machine model, w j For the weights of the assigned convolutional neural network model, set the weight threshold Y. When assigning weights, set |w z -w j |<=Y and w z +w j =1; by multiplying the failure probability of each model by its corresponding weight, the weighted failure probability is obtained, and the final failure type, failure location, failure characteristics and corresponding failure probability are output and transmitted to the fault assessment module.

9. The multi-sensor based power tool data analysis and detection method according to claim 6, characterized in that: The specific steps of step S400 for quantitatively evaluating the diagnosed fault and setting a graded warning to remind are as follows: Step S401: performing quantitative evaluation on the diagnostic results output by the fault diagnosis module, including vibration amplitude evaluation, temperature anomaly evaluation, current fluctuation evaluation and comprehensive evaluation, to further determine the fault risk level of the power tool; The vibration amplitude assessment is used to assess whether the vibration is beyond the normal range, according to the formula: VI=A v -THE vt / THE vmax -THE vt ; Where VI represents the vibration excess index, A v Indicates the vibration amplitude obtained by the sensor, A vt Indicates the normal threshold of vibration amplitude, A vmax Indicates the safe upper limit of vibration amplitude; The temperature anomaly assessment is used to assess whether the temperature of the power tool is abnormal, according to the formula: TI=|TT p | / T max -T min ; Among them, TI represents the temperature anomaly index, T represents the temperature value obtained by the sensor, Tp represents the optimal temperature for the power tool to operate, and T max and T min Respectively represent the maximum and minimum temperatures allowed under normal operating conditions; The current fluctuation estimation is used to evaluate the stability of the current. First, the average value of the current is calculated according to the formula: in represents the average value of the current, N is the number of sampling points, and In represents the current value of the nth sampling point; The standard deviation σ of the current is calculated according to the formula: Calculate the current fluctuation rate IC according to the formula: The comprehensive evaluation is to conduct a comprehensive evaluation of the power tool based on the values ​​of the vibration amplitude evaluation, the temperature anomaly evaluation, and the current fluctuation evaluation, according to the formula: F=w VI *VI+*w TI *TI+w IC *IC; where w VI 、w TI 、w IC They are the weights of vibration excess index, temperature anomaly index, and current fluctuation rate; Step S402: Based on the evaluation result of the fault evaluation unit and the output of the fault analysis module, the fault risk of the power tool is divided into different levels, and timely reminders are given through sound and light alarms and display screen prompts when the evaluation result F reaches or exceeds the preset level threshold range. At the same time, the information of each warning is recorded, including the fault type, fault characteristics, fault probability, and fault location.

Images