Concrete setting time detection device

Abstract

The invention discloses a concrete setting time detection device, which relates to the technical field of concrete detection, and is characterized in that a lifting part, an operator and a detector are linked through a closed-loop control algorithm; the closed-loop control algorithm is based on a real-time feedback mechanism, and the action of the lifting component and the control strategy of the operator are dynamically adjusted according to the working state and detection data of the detector; and a concrete setting time judgment algorithm is built in the operator and is used for judging the setting time of the concrete. A setting time judgment algorithm is built in the manipulator, multi-dimensional characteristic data of dynamic pressure, penetration resistance change rate and temperature and humidity are fused, a dual-output regression model is constructed to realize automatic setting time judgment, and the model can be updated in real time according to a concrete mix proportion or environmental change in cooperation with an incremental SVM model and a concept drift detection module, so that the accuracy of setting time judgment is improved. And the construction risk caused by misjudgment of a single index is effectively avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of concrete setting time detection, and particularly to a concrete setting time detection device. Background Art

[0002] In the field of construction engineering, concrete, as a widely used building material, its setting time is a crucial parameter, which directly affects the construction progress, construction quality, and structural safety. Accurately detecting the setting time of concrete is of great significance for reasonably arranging construction processes, ensuring the strength development and durability of concrete structures.

[0003] With the development of technology, some automated detection devices have gradually emerged. However, most of these devices have only been improved in mechanical structure to achieve partial automation of operations, but there are still great limitations in data processing and analysis. They usually adopt a simple single-threshold determination method and cannot fully consider the complex non-linear relationships and dynamic changes during the concrete setting process.

[0004] In addition, the existing detection devices also have deficiencies in anti-interference ability. The construction site environment is complex, with various noise and interference factors, such as vibration, electromagnetic interference, and environmental factors. These factors will interfere with the detection data and affect the accuracy of the detection results. The existing devices lack effective anti-interference measures and cannot effectively suppress and process these interferences. Therefore, it is necessary to propose a concrete setting time detection device. Summary of the Invention

[0005] The present invention aims to provide a concrete setting time detection device, which realizes precise control of the detector, multi-source data fusion analysis, and automatic determination of the setting time through the integration of mechanical structure innovation and intelligent algorithms, and solves the problems of low efficiency, poor accuracy, and weak adaptability of traditional methods.

[0006] To achieve the above object, the present invention adopts the following technical solution: A concrete setting time detection device includes a base. An installation seat is provided on the top of the base. A lifting component is provided inside the installation seat. An operator is provided on one side of the installation seat. A placement plate is provided on the base. A connecting plate is fixedly connected to the lifting component. A connecting sleeve is screwed onto the connecting plate. A detector is sleeved inside the inner wall of the connecting sleeve. The lifting component, the operator, and the detector are linked through a closed-loop control algorithm. The closed-loop control algorithm is based on a real-time feedback mechanism and dynamically adjusts the actions of the lifting component and the control strategy of the operator according to the working state and detection data of the detector. The operator is built-in with a concrete setting time determination algorithm for determining the setting time of concrete.

[0007] Preferably, the lifting component includes a lead screw disposed in the mounting base. A motor is installed on the top of the mounting base, and the output end of the motor is coaxially connected to the lead screw; Among them, the motor is a servo motor, and the operator controls the motor based on the fuzzy adaptive PID control algorithm. The specific formula of the fuzzy adaptive PID control algorithm is as follows: ; Among them represents the actual rotation angle of the motor at the current time t, represents the rotation angle value that the motor should reach as preset in the detection process, represents the deviation between the actual rotation angle and the target rotation angle of the motor, which is recorded as the rotation angle error, that is , respectively represent the proportional coefficient, integral coefficient, and differential coefficient at time t, represents the rotation angle error The rate of change with time is recorded as the rotation angle change rate; The rotation angle error and the rotation angle change rate are each divided into seven fuzzy subsets to construct a fuzzy rule base for setting the motor. There are forty-nine fuzzy rules, and each rule matches a value; The rotation angle error and the rotation angle change rate are obtained in real time to match the fuzzy rules and output the corresponding value.

[0008] Preferably, a moving block is threadedly connected to the lead screw. The moving block is fixedly connected to the connecting plate by bolts. One end of the connecting plate is arc-shaped, and a sliding hole is provided in the middle of the connecting plate. The connecting plate slides on the guide rod, and the sliding fit between the connecting plate and the guide rod adopts a linear bearing structure; The operator uses the laser ranging calibration algorithm based on Kalman filtering to calibrate the sliding fit between the connecting plate and the guide rod; The calibration formula of the laser ranging calibration algorithm based on Kalman filtering is: ; Among them is the estimated value of the clearance after calibration at the kth moment, is the estimated value of the predicted clearance at the kth moment, is the Kalman gain, is the actual measured clearance value obtained by laser ranging at the kth moment, and H is the measurement matrix.

[0009] Preferably, a support plate is welded to one side of the top of the mounting base, and a guide rod is fixedly connected to the top of the support plate. One end of the guide rod is fixed on the base.

[0010] Preferably, the side wall of the connecting sleeve is provided with an abutment column for fixing the detector.

[0011] Preferably, the placement plate adopts a three-stage shock-absorbing structure, including an anti-slip pad, a cache foot pad and a spring damper.

[0012] Preferably, the detector has a built-in dual-axis acceleration sensor for real-time monitoring of the vibration of the detector in the X-axis and Y-axis directions; the dual-axis acceleration sensor uses a vibration suppression algorithm based on wavelet transform and adaptive control to suppress the influence of vibration on the detection results, and the specific steps are as follows: Obtain the vibration signal a(t) collected by the dual-axis acceleration sensor and perform wavelet transform to decompose the signal into sub-signals of different frequencies: ; in is the wavelet coefficient, is the wavelet basis function, , j represents the scale of wavelet decomposition, J represents the maximum scale of decomposition, k represents the translation in time, and K represents the maximum number of translations at each scale; the frequency characteristics of the vibration signal are analyzed according to the wavelet coefficients, and the main vibration frequency components are identified; when the vibration frequency is greater than its corresponding vibration threshold, the adaptive control strategy is started; the adaptive control algorithm is used to dynamically adjust the operation mode of the motor according to the frequency and amplitude of the vibration, so that the motor automatically switches to a preset low-speed mode.

[0013] Preferably, the operator has a built-in concrete setting time determination algorithm, and the specific steps of the concrete setting time determination algorithm are as follows: Get the dynamic pressure value collected by the detector , penetration resistance change rate , Biaxial vibration acceleration signal , Ambient temperature and humidity , forming the feature vector , initial setting time measured by standard penetration method and final setting time To supervise the labels, a dual regression model of initial setting / final setting time is constructed: S1, obtaining historical detection data of concrete setting, including concrete of different grades and corresponding environmental information; S2, using historical detection data to build a training set and optimizing the PSO-hybrid kernel SVM parameters through five-fold cross validation; S3, establish the initial setting / final setting time double regression model and output the continuous prediction value , , and scored by confidence , evaluate the prediction reliability, and the confidence score C is calculated based on the support vector spacing and the classification margin; S4. Obtain the feature vectors collected by the detector in real time and input them into the initial / final setting time double regression model, and synchronously calculate the SHAP value to monitor feature anomalies; When the predicted confidence C is greater than its preset confidence threshold, the final setting time determination result is output; if the predicted confidence C is less than or equal to its preset confidence threshold, data re - collection and model incremental learning are automatically triggered.

[0014] Preferably, in the concrete setting time determination algorithm, an improved hybrid kernel function is adopted , where is the weight coefficient, which is dynamically adjusted by the particle swarm optimization algorithm and the kernel parameters. The kernel parameters include of the RBF and d of the polynomial kernel; Then introduce the fuzzy membership function to adjust the sample weights and suppress the interference of noise data on the model. The weight update formula is: ; where g is the slope parameter, is the error threshold.

[0015] Preferably, in the concrete setting time determination algorithm, a dynamic learning mechanism is also set up. The dynamic learning mechanism includes: An incremental SVM model construction module, which is used to construct an incremental SVM model, and uses the sliding window technology to import new detection data in real time to trigger the model online update mechanism; when the number of new samples reaches the window capacity N, the support vector set is updated by the fast incremental learning algorithm; A concept drift detection module, which monitors the distribution of model prediction residuals based on the statistical process method, and automatically triggers model reconstruction when a significant change in the data distribution is detected.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By integrating the multi - dimensional feature data of dynamic pressure, penetration resistance change rate, temperature and humidity through the built - in setting time determination algorithm in the operator, the present invention constructs a double - output regression model to realize automatic setting time determination. And with the cooperation of the incremental SVM model and the concept drift detection module, the model can be updated in real time according to the concrete mix ratio or environmental changes, effectively avoiding construction risks caused by misjudgment of a single index. Brief Description of the Drawings

[0017] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and the schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 Schematic diagram of the overall first - perspective structure proposed by the present invention; Figure 2 Schematic diagram of the second - perspective structure proposed by the present invention; Figure 3 Schematic diagram of some parts proposed by the present invention; Figure 4 Schematic diagram of the detector structure proposed by the present invention; Figure 5 Principle block diagram of the concrete setting time determination algorithm proposed by the present invention.

[0018] Numbers in the figure: 1, base; 2, mounting seat; 3, operator; 4, motor; 5, placing plate; 6, lead screw; 7, moving block; 8, connecting plate; 9, support plate; 10, guide rod; 11, connecting sleeve; 12, detector; 13, abutting column. Detailed implementation manners

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0020] See Figures 1 - 5 , a concrete setting time detection device in the present invention includes a base 1. A mounting seat 2 is provided on the top of the base 1, and the mounting seat 2 is used to carry and fix the core components of the device. A lifting component is provided in the mounting seat 2, and its function is to accurately control the vertical position of the detector 12. An operator 3 is provided on one side of the mounting seat 2. The operator 3 is the control and data - processing center of the whole device, responsible for parameter setting, process control, and data analysis and storage; a placing plate 5 is provided on the base 1 for placing the concrete sample to be detected. A connecting plate 8 is fixedly connected to the lifting component. When the lifting component moves, the connecting plate 8 moves synchronously. A connecting sleeve 11 is screwed on the connecting plate 8, and the detector 12 is sleeved on the inner wall of the connecting sleeve 11. Through this connection method, the stable installation and flexible replacement of the detector 12 are ensured; The lifting component, the operator 3, and the detector 12 are linked through a closed - loop control algorithm; the closed - loop control algorithm is based on a real - time feedback mechanism. According to the working state and detection data of the detector 12, the actions of the lifting component and the control strategy of the operator 3 are dynamically adjusted to ensure the stability and reliability of the detection process; the operator 3 is built - in with a concrete setting time determination algorithm for determining the setting time of the concrete.

[0021] In the present invention, the lifting component includes a lead screw 6 provided in the mounting seat 2. A motor 4 is installed on the top of the mounting seat 2, and the output end of the motor 4 is coaxially connected to the lead screw 6; Among them, the motor 4 is a servo motor, and the operator 3 controls the motor 4 based on the fuzzy adaptive PID control algorithm; the specific formula of the fuzzy adaptive PID control algorithm is as follows: ; Among them represents the actual rotation angle of the motor at the current moment t, represents the rotation angle value that the motor should reach as preset in the detection process, represents the deviation between the actual rotation angle and the target rotation angle of the motor, denoted as the rotation angle error, that is , respectively represent the proportional coefficient, integral coefficient, and differential coefficient at the moment t, represents the rotation angle error The rate of change with time is denoted as the rotation angle change rate; The rotation angle error and the rotation angle change rate are each divided into seven fuzzy subsets to construct a fuzzy rule base for setting the motor 4. Suppose there are forty-nine fuzzy rules, and each rule matches a value; The rotation angle error and the rotation angle change rate are obtained in real time to match the fuzzy rules, and the corresponding value is output.

[0022] In the present invention, a moving block 7 is threadedly connected to the lead screw 6. The moving block 7 is fixedly connected to the connecting plate 8 by bolts. One end of the connecting plate 8 is arc-shaped. A sliding hole is provided in the middle section of the connecting plate 8. The connecting plate 8 slides on the guide rod 10. The sliding fit between the connecting plate 8 and the guide rod 10 adopts a linear bearing structure; the operator 3 uses a laser ranging calibration algorithm based on Kalman filtering to calibrate the sliding fit between the connecting plate 8 and the guide rod 10; The calibration formula of the laser ranging calibration algorithm based on Kalman filtering is: ; Among them is the estimated value of the clearance after calibration at the k-th moment, is the predicted estimated value of the clearance at the k-th moment, is the Kalman gain, is the actual measured clearance value obtained by laser ranging at the k-th moment, and H is the measurement matrix.

[0023] It should be noted that the Kalman filtering algorithm can effectively fuse historical measurement data and current measurement data, filter the measurement noise, and improve the accuracy of measurement. Through multiple samplings and calibration using this algorithm, the clearance error between the connecting plate 8 and the guide rod 10 is ensured, thus ensuring the smoothness and accuracy of the sliding.

[0024] In the present invention, one side of the top of the mounting base 2 is welded with a support plate 9, the top of the support plate 9 is fixedly connected with a guide rod 10, and one end of the guide rod 10 is fixed on the base 1.

[0025] In the present invention, the side wall of the connecting sleeve 11 is provided with an abutting column 13 for fixing the detector 12.

[0026] In the present invention, the placing plate 5 adopts a three-stage shock absorption structure, including an anti-slip pad, a buffer foot pad and a spring damper.

[0027] It should be noted that the anti-slip pad can increase the friction between the placing plate 5 and the base 1 to prevent the placing plate 5 from sliding during the detection process; the buffer foot pad can initially absorb the vibration energy transmitted from the outside; and the spring damper can further attenuate and buffer the vibration.

[0028] In the present invention, the detector 12 is internally provided with a biaxial acceleration sensor for real-time monitoring of the vibration conditions of the detector 12 in the X-axis and Y-axis directions; the biaxial acceleration sensor adopts a vibration suppression algorithm based on wavelet transform and adaptive control to suppress the influence of vibration on the detection result. The specific steps are as follows:

[0029] Acquire the vibration signal a(t) collected by the biaxial acceleration sensor and perform wavelet transform to decompose the signal into sub-signals of different frequencies: ; where is the wavelet coefficient, is the wavelet basis function, , j represents the scale of wavelet decomposition, J represents the maximum scale of decomposition, k represents the translation in time, and K represents the maximum number of translations at each scale; analyze the frequency characteristics of the vibration signal according to the wavelet coefficient to identify the main vibration frequency components; when the vibration frequency is greater than its corresponding vibration threshold, start the adaptive control strategy; the adaptive control algorithm is used to dynamically adjust the operation mode of the motor 4 according to the frequency and amplitude of the vibration, so that the motor 4 automatically switches to the preset low-speed mode.

[0030] In the present invention, the operator 3 is internally provided with a concrete setting time determination algorithm. The specific steps of its concrete setting time determination algorithm are as follows: Obtain the dynamic pressure value collected by the detector 12 , the change rate of penetration resistance , the biaxial vibration acceleration signal , the ambient temperature and humidity , to form a feature vector , the initial setting time and the final setting time For the supervision label, a double regression model of initial setting / final setting time is constructed: S1. Obtain the historical detection data of concrete setting, including different grades of concrete and corresponding environmental information; S2. Use the historical detection data to construct a training set, and optimize the PSO - hybrid kernel SVM parameters through five - fold cross - validation; S3. Establish a double regression model of initial setting / final setting time and output continuous prediction values 、 and evaluate the prediction reliability through a confidence score . The confidence score C is calculated based on the support vector distance and classification margin; S4. Obtain the feature vectors real - time collected by the detector 12 and input them into the double regression model of initial setting / final setting time, and synchronously calculate the SHAP value to monitor feature anomalies; When the prediction confidence C is greater than its preset confidence threshold, the final setting time determination result is output; if the prediction confidence C is less than or equal to its preset confidence threshold, data re - collection and model incremental learning are automatically triggered.

[0031] In the present invention, in the concrete setting time determination algorithm, an improved hybrid kernel function is adopted, where is the weight coefficient, which is dynamically adjusted by the particle swarm optimization algorithm and the kernel parameters. The kernel parameters include of RBF and d of the polynomial kernel, which improves the fitting ability of the model to non - linear data; Then, a fuzzy membership function is introduced to adjust the sample weights and suppress the interference of noise data on the model. The weight update formula is: ; where g is the slope parameter, is the error threshold, and the model robustness is improved by adaptively adjusting the sample weights.

[0032] In the present invention, in the concrete setting time determination algorithm, a dynamic learning mechanism is also set up. The dynamic learning mechanism includes: An incremental SVM model construction module, which is used to construct an incremental SVM model, uses the sliding window technology to import new detection data in real - time and trigger the model online update mechanism; when the number of new samples reaches the window capacity N, the support vector set is updated through the fast incremental learning algorithm; A concept drift detection module, which monitors the distribution of model prediction residuals based on the statistical process method. When a significant change in the data distribution is detected, the model reconstruction is automatically triggered; The determination logic of the significant change is: Obtain the current concrete mix ratio, calculate the similarity between the current concrete mix ratio and the previous concrete mix ratio to obtain the mix ratio similarity; if the mix ratio similarity is less than the mix ratio change threshold, it is determined that the mix ratio has changed significantly. Obtain the current environmental factors, including temperature, humidity, and vibration amplitude; set the standard values of the environmental factors, and subtract the corresponding value of the environmental factor from the standard value of any environmental factor to obtain the factor difference corresponding to the environmental factor. Perform normalized weighted calculation on the factor differences of all environmental factors to obtain the environmental change value; if the environmental change value is greater than the preset environmental change threshold, it is determined that the environment has changed significantly; record the significant changes in the mix ratio and the environment as significant changes.

[0033] As described above, only the preferred specific embodiments of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A concrete setting time detection device, comprising a base (1), characterized in that: A mounting seat (2) is provided on the top of the base (1), a lifting component is provided inside the mounting seat (2), an operator (3) is provided on one side of the mounting seat (2), a placement plate (5) is provided on the base (1), a connecting plate (8) is fixedly connected to the lifting component, a connecting sleeve (11) is screwed on the connecting plate (8), and a detector (12) is sleeved on the inner wall of the connecting sleeve (11); The lifting component, the operator (3) and the detector (12) are linked via a closed-loop control algorithm; the closed-loop control algorithm is based on a real-time feedback mechanism and dynamically adjusts the action of the lifting component and the control strategy of the operator (3) according to the working state and detection data of the detector (12); the operator (3) has a built-in concrete setting time determination algorithm for determining the setting time of concrete.

2. A concrete setting time detection device according to claim 1, characterized in that: The lifting component comprises a screw rod (6) arranged in a mounting seat (2); a motor (4) is mounted on the top of the mounting seat (2); an output end of the motor (4) is coaxially connected to the screw rod (6); The motor (4) is a servo motor, and the operator (3) controls the motor (4) based on a fuzzy adaptive PID control algorithm; the specific formula of the fuzzy adaptive PID control algorithm is as follows: ； in Indicates the actual rotation angle of the motor at the current time t, Indicates the rotation angle value that the motor should reach in advance according to the detection process. The deviation between the actual rotation angle of the motor and the target rotation angle is recorded as the rotation angle error, that is, , They represent the proportional coefficient, integral coefficient and differential coefficient at time t respectively. Indicates the angle error The rate of change over time is the rate of change of the angle of the Discipline Inspection Commission; The rotation angle error and the rate of change of the rotation angle Each of them is divided into seven fuzzy subsets, and a fuzzy rule base for setting motor (4) is constructed. There are 49 fuzzy rules, each of which matches a The value of; Get the angle error in real time and the rate of change of the rotation angle Match the fuzzy rules and output the corresponding The value of .

3. A concrete setting time detection device according to claim 1, characterized in that: The screw rod (6) is threadedly connected with a moving block (7), the moving block (7) is fixedly connected to a connecting plate (8) by means of bolts, one end of the connecting plate (8) is in an arc shape, a sliding hole is provided in the middle of the connecting plate (8), the connecting plate (8) slides on a guide rod (10), and the sliding fit between the connecting plate (8) and the guide rod (10) adopts a linear bearing structure; the operator (3) uses a laser ranging calibration algorithm based on Kalman filtering to calibrate the sliding fit of the connecting plate (8) on the guide rod (10); wherein the calibration formula of the laser ranging calibration algorithm based on Kalman filtering is: ;in is the estimated value of the fit clearance after calibration at the kth moment, is the estimated value of the fit clearance predicted at the kth moment, is the Kalman gain, is the actual measurement gap value obtained by laser ranging at the kth moment, and H is the measurement matrix.

4. A concrete setting time detection device according to claim 1, characterized in that: A support plate (9) is welded to one side of the top of the mounting seat (2), a guide rod (10) is fixedly connected to the top of the support plate (9), and one end of the guide rod (10) is fixed to the base (1).

5. A concrete setting time detection device according to claim 1, characterized in that: The side wall of the connecting sleeve (11) is provided with an abutment column (13) for fixing the detector (12).

6. A concrete setting time detection device according to claim 1, characterized in that: The placement plate (5) adopts a three-stage shock-absorbing structure, including an anti-slip pad, a buffer pad and a spring damper.

7. A concrete setting time detection device according to claim 1, characterized in that: The detector (12) has a built-in dual-axis acceleration sensor for real-time monitoring of the vibration of the detector (12) in the X-axis and Y-axis directions; the dual-axis acceleration sensor uses a vibration suppression algorithm based on wavelet transform and adaptive control to suppress the influence of vibration on the detection result, and the specific steps are as follows: Obtain the vibration signal a(t) collected by the dual-axis acceleration sensor and perform wavelet transform to decompose the signal into sub-signals of different frequencies: ； in is the wavelet coefficient, is the wavelet basis function, , j represents the scale of wavelet decomposition, J represents the maximum scale of decomposition, k represents the translation in time, and K represents the maximum number of translations at each scale; the frequency characteristics of the vibration signal are analyzed according to the wavelet coefficients to identify the main vibration frequency components; when the vibration frequency is greater than its corresponding vibration threshold, the adaptive control strategy is started; the adaptive control algorithm is used to dynamically adjust the operation mode of the motor (4) according to the frequency and amplitude of the vibration, so that the motor (4) automatically switches to a preset low-speed mode.

8. A concrete setting time detection device according to claim 1, characterized in that: The operator (3) has a built-in concrete setting time determination algorithm, and the specific steps of the concrete setting time determination algorithm are as follows: Obtain the dynamic pressure value collected by the detector (12) , penetration resistance change rate , Biaxial vibration acceleration signal , Ambient temperature and humidity , forming the feature vector , initial setting time measured by standard penetration method and final setting time To supervise the labels, a dual regression model of initial setting / final setting time is constructed: S1, obtaining historical detection data of concrete setting, including concrete of different grades and corresponding environmental information; S2, using historical detection data to build a training set and optimizing the PSO-hybrid kernel SVM parameters through five-fold cross validation; S3, establish the initial setting / final setting time double regression model and output the continuous prediction value , , and scored by confidence , evaluate the prediction reliability, the confidence score C is calculated based on the support vector spacing and classification interval; S4, obtaining the feature vector collected by the detector (12) in real time and inputting it into the initial setting / final setting time double regression model, and synchronously calculating the SHAP value to monitor feature anomalies; When the prediction confidence C is greater than its preset confidence threshold, the final coagulation time determination result is output; if the prediction confidence C is less than or equal to its preset confidence threshold, data re-collection and model incremental learning are automatically triggered.

9. A concrete setting time detection device according to claim 8, characterized in that: In the concrete setting time determination algorithm, an improved hybrid kernel function is used. ,in is the weight coefficient, which is dynamically adjusted by the particle swarm optimization algorithm And kernel parameters, kernel parameters include RBF , d of the polynomial kernel; Then introduce the fuzzy membership function The sample weights are adjusted to suppress the interference of noise data on the model. The weight update formula is: ; where g is the slope parameter, is the error threshold.

10. A concrete setting time detection device according to claim 8, characterized in that: In the concrete setting time determination algorithm, a dynamic learning mechanism is also provided, and the dynamic learning mechanism includes: The incremental SVM model building module is used to build an incremental SVM model. It uses the sliding window technology to import new detection data in real time and trigger the model online update mechanism. When the number of new samples reaches the window capacity N, the support vector set is updated through a fast incremental learning algorithm. The concept drift detection module monitors the distribution of model prediction residuals based on statistical process methods and automatically triggers model reconstruction when significant changes in data distribution are detected.

Images