Intelligent cabinet door anti-pinch control method and system based on multi-source sensing fusion

Abstract

The invention relates to the technical field of automatic control and data processing, in particular to an intelligent cabinet door anti-pinch control method and system based on multi-source sensing fusion, and the method comprises the steps: obtaining the current and speed data in the closing process of an intelligent cabinet door, and determining the resistance drift consistency degree in combination with factory reference data; analyzing the high-frequency oscillation energy ratio of the current data and the change trend correlation between the current and the speed, and comprehensively constructing an environmental state confidence coefficient; and calculating a self-adaptive inertia weight of a particle of the PSO algorithm based on the environmental state confidence so as to drive the PSO algorithm to perform parameter optimization on the SVM classification model, so that the classification model dynamically adapts to the current environment, an optimized SVM model is obtained, and intelligent anti-pinch control is performed on the cabinet door. According to the method, an environment perception and self-adaptive optimization mechanism is introduced, so that resistance abnormity caused by environmental factors and a real person / cargo clamping state can be accurately distinguished, and the accuracy of anti-pinch control is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic control and data processing. Specifically, it relates to an intelligent cabinet door anti-pinch control method and system based on multi-source sensor fusion. BACKGROUND

[0002] With the vigorous development of the new retail industry, intelligent vending cabinets, as key terminals connecting online and offline, are widely used in community, office building and school scenes. Considering that such equipment is usually unattended and open, and the user group covers a wide range, including the elderly and children, the safety, sensitivity and reliability of the cabinet door anti-pinch control function are particularly important, and are directly related to the personal safety of users and the operation efficiency of the equipment.

[0003] In the prior art, the control of the intelligent cabinet door anti-pinch usually relies on real-time monitoring and classification of motor operation data. The mainstream solution is to use an SVM (Support Vector Machine) classification model to identify normal door closing states and abnormal pinch states by collecting current, speed and other characteristic data during the motor operation of the cabinet door. In order to further improve the classification accuracy of SVM in nonlinear and high-dimensional data, the traditional method introduces a PSO (Particle Swarm Optimization) algorithm to optimize the key parameters (penalty factor and kernel function parameter) in the SVM model. The method uses the iterative search strategy of swarm intelligence to overcome the blindness of manually setting parameters to some extent, and improves the generalization ability and classification accuracy of the model.

[0004] However, intelligent vending cabinets are faced with outdoor or semi-outdoor environmental conditions, especially in low temperature seasons or in the case of equipment aging. The door sealing strip will harden due to temperature reduction, or the guide rail will stick due to oil accumulation. These environmental factors will cause the motor running resistance to gradually drift as a whole or to produce micro-stick oscillation. Under such complex time-varying working conditions, the characteristic distribution of the data during the motor operation will shift, causing the high resistance samples that belong to normal door closing to overlap with abnormal sample distribution, and thus being incorrectly classified as an abnormal category.

[0005] In the face of the above working condition changes, the traditional PSO algorithm usually adopts a fixed or simply linearly decreasing inertia weight when performing parameter optimization. This rigid search strategy lacks the ability to perceive and respond to real-time environmental conditions. When the environmental resistance drifts, the system urgently needs the algorithm to perform a rapid local fine search near the current working condition parameters to quickly update the SVM parameters to adapt to the environment. However, due to the inability of the inertia weight to adaptively adjust according to the environmental confidence, the algorithm still tends to maintain a large range of global search capability, causing the PSO algorithm to fail to converge quickly to the local optimal solution that adapts to the current high-resistance working condition within a limited control period, resulting in the SVM model failing to adjust the classification boundary in time. The normal closing resistance affected by the hardening of the low-temperature adhesive tape or oil stains is incorrectly determined as a pinch abnormal signal. This high-frequency misjudgment will cause the cabinet door to frequently trigger the anti-pinch rebound without any obstacles, which cannot be normally closed. Not only does this destroy the refrigeration and insulation effect of the intelligent cabinet, increase the energy consumption of the equipment, but also cause the equipment to frequently fail and stop operating, affecting the shopping experience of users. SUMMARY

[0006] To solve the problem of the existing PSO algorithm and SVM classification model based anti-pinch control method, which lacks environmental adaptive ability when facing complex working conditions such as low temperature or oil stains, resulting in misjudgment of normal closing high resistance as pinch abnormality, and further causing frequent misrebound of the cabinet door, the present application provides an intelligent cabinet door anti-pinch control method and system based on multi-source sensor fusion.

[0007] In a first aspect, the present application provides an intelligent cabinet door anti-pinch control method based on multi-source sensor fusion, comprising: Obtaining the current data sequence and the speed data sequence of the intelligent cabinet door during each start-to-stop closing process, and the mean value of the current data of multiple cabinet door start-to-stop closing processes tested under the factory standard state of the intelligent cabinet door, to form a reference current data sequence; Based on the numerical difference between the current data sequence and the reference current data sequence, determining the consistency degree of resistance drift of each start-to-stop closing process; Extracting the proportion of high-frequency oscillation energy of the current data sequence, calculating the correlation of the change trend between the current data sequence and the speed data sequence, and comprehensively determining the environmental state confidence of each start-to-stop closing process based on the consistency degree of resistance drift, the correlation of the change trend, and the proportion of high-frequency oscillation energy; Based on the environmental state confidence, the adaptive inertia weight of the PSO algorithm is calculated in each iteration. The adaptive inertia weight is used to optimize the parameters of the SVM classification model. The output of the optimized SVM classification model is used to determine whether the state of the cabinet door from the start of closing to the stop of closing is normal or abnormal. The intelligent cabinet door is then controlled to prevent pinching based on the state.

[0008] This technical solution reflects the long-term benchmark shift caused by mechanical aging by introducing resistance drift consistency. It comprehensively evaluates the reliability of data collected under the working conditions of the smart vending machine by using environmental state confidence. It also establishes a dynamic parameter optimization mechanism. When the environmental confidence is high, it indicates that the data conforms to physical laws and has low noise. The inertial weight of particles in the PSO algorithm is adjusted in the parameter optimization process, so that the SVM classification model can accurately fit the characteristics of the current data sequence and velocity data sequence of the cabinet door from the start of closing to the stop of closing. When the environment is complex, the strategy is adaptively adjusted to ensure robustness. This closed-loop control based on environmental perception and model adaptive optimization avoids the interference of mechanical wear and environmental vibration on the anti-pinch control judgment and improves the accuracy of cabinet door anti-pinch control.

[0009] Preferably, the degree of uniformity of drag drift is determined based on the following method: for the first Alignment is performed on the current and velocity data sequences from the start of closure to the end of closure; drag drift consistency is calculated.

[0010] in, For the first The degree of consistency of resistance drift during the entire closing process from the start to the end of the closing process. For the first The length of the current data sequence from the start of the closing process to the end of the closing process. For the first The current data sequence from the start of closing the secondary cabinet door to the stop of closing. A number, The first in the reference current data sequence A number, For the first The overall offset between the current data sequence from the start of closing of the secondary cabinet door to the stop of closing and the reference current data sequence, wherein the overall offset is the total offset of the current data sequence from the start of closing of the secondary cabinet door to the stop of closing. The current data sequence from the start of the closing process to the stop of the closing process is obtained by subtracting the reference current data sequence element by element and then averaging the results. To prevent constants with a denominator of zero, This is a normalization operation.

[0011] The technical scheme precisely reflects the shape characteristics of the resistance change instead of the numerical value size on a mathematical level by constructing a nonlinear consistency evaluation model based on residual dispersion, when the intelligent cabinet is in a low-temperature or aging state, the running resistance shows a uniform overall drift in the whole stroke, and the difference fluctuation is extremely small, so that a very high consistency index is calculated, and when an emergency such as pinch occurs, the resistance curve will produce local distortion, resulting in a sharp increase in difference dispersion, and this processing method effectively separates the systematic environmental drift and random obstacle mutation from the physical origin.

[0012] Preferably, the method for extracting the proportion of high-frequency oscillation energy of the current data sequence is: the sum of squares of all values in the current data sequence is taken as the total energy of the current data sequence; the discrete wavelet transform technology is used to perform multi-scale decomposition on the current data sequence to obtain the high-frequency detail coefficient sequence under each scale decomposition; the sum of squares of all values in the high-frequency detail coefficient sequence under the minimum scale decomposition is taken as the high-frequency oscillation energy; and the ratio of the high-frequency oscillation energy to the total energy is determined as the proportion of the high-frequency oscillation energy of the current data sequence.

[0013] Preferably, the change trend correlation between the current data sequence and the speed data sequence is determined based on the following manner: first-order difference operation is performed on the current data sequence and the speed data sequence respectively to obtain two first-order difference sequences, and the Pearson correlation coefficient between the two first-order difference sequences is taken as the change trend correlation between the current data sequence and the speed data sequence.

[0014] Preferably, the environmental state confidence of each start-closing-to-stop-closing process satisfies the following relationship:

[0015] wherein, is the environmental state confidence of the n th start-closing-to-stop-closing process, is a normalization operation, is the resistance drift consistency degree of the n th start-closing-to-stop-closing process, is the high-frequency oscillation energy of the current data sequence of the n th start-closing-to-stop-closing process, is the total energy of the current data sequence of the n th start-closing-to-stop-closing process, is the change trend correlation between the current data sequence and the speed data sequence of the n th start-closing-to-stop-closing process, is the absolute value symbol.

[0016] ​​​​​This technical solution constructs a multi-dimensional feature fusion evaluation system. It captures the overall macroscopic resistance increase caused by low temperature by utilizing the consistency of resistance drift, and captures the microscopic mechanical oscillation caused by oil viscosity by utilizing the correlation between the proportion of high-frequency oscillation energy and the trend of change. Through nonlinear weighted fusion, it can accurately separate the essential difference between nonlinear resistance caused by environmental factors and abrupt resistance caused by obstacle collision from the physical level. It ensures that high confidence is output only when the signal characteristics highly conform to the laws of mechanical aging or environmental drift, thus providing a judgment benchmark with extremely strong anti-interference ability for subsequent control algorithms.

[0017] Preferably, the adaptive inertia weights of the PSO algorithm in each iteration satisfy the following relationship:

[0018] in, In order to target the The PSO algorithm performs the parameter optimization process for the SVM classification model from the start of closure to the stop of closure. The adaptive inertia weights of all particles in the next iteration. and These are the maximum and minimum inertia weights preset for the PSO algorithm, respectively. For the first Confidence of environmental states during the process from the start of closure to the end of closure.

[0019] This technical solution establishes a dynamic switching mechanism for particle swarm search strategies based on environmental awareness. When the confidence level of the environmental state is high, it indicates that the current high drag originates from environmental factors. The algorithm automatically reduces the inertial weight to suppress the global divergence of particles, prompting them to conduct a local fine-grained search in the solution space near the current operating condition. This allows the algorithm to quickly converge to new parameters that fit the current environment within a millisecond-level control cycle. Conversely, when the confidence level is low, the algorithm maintains a large inertial weight to preserve global search vitality and prevent the model from overfitting to sudden signals. This dynamic adjustment achieves a balance between rapid adaptation under steady-state conditions and high-sensitivity response under sudden operating conditions.

[0020] Preferably, the method for parameter optimization of the SVM classification model by using the adaptive inertia weight is: constructing a training sample library, including: a plurality of normal sample data in a factory standard state, a plurality of abnormal sample data when the cabinet door produces a person / goods, and a plurality of normal sample data recently judged as the source of resistance change being environmental factors, and the cabinet door being in a normal closed state, wherein the normal / abnormal sample data in the training sample library each contains a current data sequence and a speed data sequence of a process of starting to close to stopping to close of the cabinet door; defining a fitness function of the PSO algorithm, representing the cross-validation average accuracy of the parameter combination represented by the particle on the training sample library; updating the particle state by using the adaptive inertia weight in the iterative optimization process of the PSO algorithm to search for the optimal parameter combination; and constructing an SVM classification model by using the optimal parameter combination when the maximum number of iterations is reached.

[0021] The technical solution uses a first-in-first-out sliding window mechanism to construct a dynamic environment sample set, introduces the latest high-resistance sample judged as an environmental factor in real time and eliminates old samples, forces the SVM classification boundary to dynamically evolve with the change of the environment, prevents the dullness of the model caused by sample lag, ensures that the model can accommodate the current harsh working conditions, and, in combination with the fitness function based on the cross-validation accuracy, ensures that the finally generated SVM model parameters can accurately identify environmental interference, so that the door closing operation is correctly performed, and the accuracy of the cabinet door anti-pinch control is improved.

[0022] Preferably, the alignment operation is performed based on the following manner: discrete time integration is performed by using the speed data sequence to obtain the cumulative displacement amount of the door body at each time in the process of starting to close to stopping to close, so that the correspondence relationship of the cumulative displacement amount of the door body, the current, and the speed at each time is constructed; a standardized door body stroke space coordinate system is set, and the door body stroke space coordinate system is divided into a plurality of equally spaced standard space sampling points; the current and the speed corresponding to each standard space sampling point are calculated by using the cumulative displacement amount of the door body as an index and applying a linear interpolation algorithm, so that the speed data sequence and the current data sequence one-to-one correspond based on the standard space sampling points, so that the alignment operation is realized.

[0023] In a second aspect, the present application also provides an intelligent cabinet door anti-pinch control system based on multi-source sensing fusion, characterized in that the intelligent cabinet door anti-pinch control system comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the steps of the intelligent cabinet door anti-pinch control method.

[0024] The present application has the following effects: The application effectively solves the false anti-pinch problem caused by resistance drift of the intelligent cabinet door under complex working conditions such as low temperature and oil stains, deeply fuses the macro drift consistency feature and the micro shock energy feature of the motor current, and can accurately identify the environmental resistance caused by non-human factors; the adaptive inertia weight strategy is used to automatically switch to the local fine search mode when the environmental disturbance is confirmed, so that the SVM model quickly reconstructs the classification boundary to contain the high resistance normal state, and the global sensitivity is maintained in the event of a sudden pinch, the control strategy combining stable state adaptation and sudden high response can significantly reduce the false detection rate of the equipment, improve the operation efficiency, strictly protect the personal safety of the user, and realize the double improvement of environmental robustness and anti-pinch reliability. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a method flowchart of the application; Figure 2 is a schematic diagram of the change trend of the cabinet door running state score and the anti-pinch trigger threshold output by the application and the traditional SVM classification model. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application.

[0027] Reference Figure 1 The application provides an intelligent cabinet door anti-pinch control method based on multi-source sensing fusion, comprising: S1: acquiring current data sequence and speed data sequence of the intelligent cabinet door in each start closing to stop closing process, and a reference current data sequence.

[0028] The cabinet door driving system of the intelligent vending cabinet needs to meet high-frequency start and stop while ensuring the absolute safety of special user groups such as the elderly and children, however, in the actual hardware architecture, in order to control the cost and adapt to the compact cabinet space, the driving motor is usually co-located with strong electrical equipment such as a refrigeration compressor, resulting in that the collected current signal often contains a large amount of electromagnetic radiation noise and mechanical commutation ripple. In addition, the Hall position sensor of the vending cabinet is limited by installation accuracy and is prone to quantization error in the low-speed closing stage, if the original data with high noise characteristics is directly input into the subsequent algorithm, it is easy to cause the drowning of weak human body signals due to noise amplification effect.

[0029] Therefore, this step not only realizes the synchronous acquisition of current and speed, but also the core task is to build a high-fidelity signal conditioning link that adapts to the hardware characteristics of the intelligent cabinet, filter out background electromagnetic noise and restore the real load change curve caused by sealing rubber strip extrusion or foreign object obstruction through time domain filtering and dynamic window interception.

[0030] The high-precision current sampling circuit on the intelligent cabinet main control board and the motor Hall sensor are used to synchronously trigger collection at a frequency of 100 Hz, which is fully considered in view of the short door closing stroke (usually only a few seconds) of the intelligent cabinet, so that the millisecond-level resistance change in the instant of contact of the adhesive strip can be captured. The system automatically identifies the process of each door from the start of closing to the stop of closing by monitoring the door locking signal and the motor back electromotive force, and eliminates the mechanical vibration data in the instant of unlocking, so as to ensure that the subsequent calculation focuses on the resistance characteristics in the closing process.

[0031] Specifically, the current data sequence and the speed (rotational speed of the motor) data sequence of the door in each process from the start of closing to the stop of closing are collected, and a one-dimensional Gaussian filter is used to smooth the original current sequence. This operation can effectively smooth the isolated peaks caused by electromagnetic interference, while the current slowly rising trend caused by the hardening of the low-temperature adhesive strip and the mechanical shock texture caused by the oil pollution of the guide rail are retained to the greatest extent, thereby laying a data foundation for subsequent accurate distinction between environmental factors and pinch / handling accidents.

[0032] Further, the mean values of the current data sequences of a plurality of processes from the start of closing to the stop of closing of the intelligent cabinet door in the standard state at the factory are obtained, to form a reference current data sequence. For example, the intelligent cabinet door is tested 10 times in the standard state at the factory, 10 current data sequences of the 10 processes are obtained, the mean values of each current data sequence are calculated, and the 10 mean values are combined to form a reference current data sequence as a reference for subsequent comparison.

[0033] S2: Based on the numerical difference between the current data sequence and the reference current data sequence, the consistent degree of resistance drift in each process from the start of closing to the stop of closing is determined.

[0034] The intelligent vending cabinet often needs to operate outdoors or in a cold chain environment. The sealing adhesive strip of the cabinet door will harden significantly under the action of low temperature. The change of the material characteristics will cause the friction resistance of the cabinet door in the entire closing stroke to present a uniform and linear rise, which directly threatens the closing tightness of the cabinet door and the food preservation effect in the cabinet. On the contrary, the pinch accident usually shows a local resistance surge. In order to accurately identify this difference from the complex current signal, this step aims to determine the consistent degree of resistance drift in each process from the start of closing to the stop of closing by comparing the parallel degree of the measured current data sequence and the reference current data sequence in form, so as to identify whether the current resistance rise belongs to systematic overall drift, which not only prevents false judgment of anti-pinch, but also ensures that the motor can output enough torque to close the cabinet door tightly under low-temperature working conditions, thereby ensuring the operation efficiency.

[0035] In one embodiment, the consistency of resistance drift during each closing process from start to stop is determined based on the following: For the The current data sequence and velocity data sequence from the start of the closing process to the stop of the closing process are aligned. Calculate drag drift consistency:

[0036] in, For the first The degree of consistency of resistance drift during the entire closing process from the start to the end of the closing process. For the first The length of the current data sequence from the start of the closing process to the end of the closing process. For the first The current data sequence from the start of closing the secondary cabinet door to the stop of closing. A number, The first in the reference current data sequence A number, For the first The overall offset between the current data sequence from the start of closing of the secondary cabinet door to the stop of closing and the reference current data sequence, wherein the overall offset is the total offset of the current data sequence from the start of closing of the secondary cabinet door to the stop of closing. The current data sequence from the start of the closing process to the stop of the closing process is obtained by subtracting the reference current data sequence element by element and then averaging the results. To prevent the constant from having a denominator of zero, the value is taken as 0.001. For normalization operations, a nonlinear mapping function (such as the Sigmoid function) is used to compress any positive number into the range (0,1].

[0037] A nonlinear evaluation model based on residual dispersion is constructed here, specifically to distinguish between environmental aging and sudden collisions in intelligent cabinets, and the overall offset. It serves as a baseline calibration, mathematically filtering out the DC resistance component caused by rubber strip hardening or mechanical wear, allowing the calculation to focus on the fluctuation pattern of the resistance, the core term. The variance of the residuals after bias correction was statistically analyzed, which physically characterizes the smoothness of the measured curve relative to the reference curve. When When the value is large, it means that the residual variance is extremely small, indicating that the measured current has only undergone an overall shift based on the reference, and the waveform details have not been distorted. This is highly consistent with the physical characteristic of the sealing strip hardening uniformly throughout its entire stroke under low temperature conditions. Conversely, if a user's finger is pinched, the resistance curve will experience a sudden change at a specific displacement point, leading to a sharp increase in residual dispersion. The value drops sharply, and this processing method realizes the accurate separation of systematic environmental drift (which needs to be closed with more effort) and random obstacle mutation (which needs to be immediately rebounded) from the mathematical origin.

[0038] In one embodiment, the alignment operation is performed based on the following manner: Discrete-time integration is performed on the speed data sequence to obtain the cumulative displacement amount of the door body at each time of the start-closing-to-stop-closing process, thereby constructing the correspondence relationship of the cumulative displacement amount of the door body, the current, and the speed at each time; a standardized door travel space coordinate system is set, and the door travel space coordinate system is divided into a plurality of equally spaced standard space sampling points; the current and the speed corresponding to each standard space sampling point are calculated by using the cumulative displacement amount of the door body as an index and applying a linear interpolation algorithm, so that the speed data sequence and the current data sequence correspond one by one based on the standard space sampling points, thereby realizing the alignment operation.

[0039] S3: Extract the proportion of high-frequency oscillation energy of the current data sequence, calculate the correlation between the change trend of the current data sequence and the speed data sequence, and determine the environmental state confidence of each start-closing-to-stop-closing process in combination with the resistance drift consistency.

[0040] After obtaining the resistance drift consistency of each start-closing-to-stop-closing process, considering that dust and grease mixed with oil will inevitably accumulate in the guide rail and lead screw during long-term operation of the intelligent vending cabinet, this semi-fluid medium will cause the door body to produce a micro alternating process of sticking and sliding during movement. This physical phenomenon will cause the current data to superimpose high-frequency sawtooth oscillations. Due to the existence of such oscillations, the difference between the current data sequence and the reference current data sequence of each start-closing-to-stop-closing process will be significantly increased, resulting in a smaller resistance drift consistency. If subsequent control is directly based on the smaller resistance drift consistency, the algorithm will mistakenly believe that there is an uncertain obstacle interference, thereby maintaining a larger inertia weight for global search, which cannot quickly adapt to the current low-temperature oil pollution working condition, and still leads to false detection.

[0041] Therefore, based on the consistency of the resistance drift, the step further analyzes that the real obstacle collision usually causes a monotonic surge in the current and a sharp drop in the speed, while the low-temperature oil pollution phenomenon is manifested as high-frequency micro-oscillation of the current, and the oscillation and the slight fluctuation of the speed are highly synchronized. Therefore, if the energy proportion of the high-frequency oscillation of the current data sequence is larger and the cross-correlation between the current oscillation and the speed oscillation is stronger, it indicates that the process of starting to close the cabinet door to stopping closing the cabinet door does not encounter an obstacle (such as pinching a person or goods), but is in a special low-temperature oil pollution working condition. At this time, the confidence degree in the environmental state should be higher, that is, the environmental state confidence degree should be larger. On the contrary, if the energy proportion of the high-frequency oscillation of the current data sequence is smaller or the cross-correlation between the current oscillation and the speed oscillation is weaker, it indicates that the process of starting to close the cabinet door to stopping closing the cabinet door encounters an obstacle, rather than being in a special low-temperature oil pollution working condition. At this time, the confidence degree in the environmental state should be lower, that is, the environmental state confidence degree should be smaller.

[0042] The step introduces micro-dimension feature analysis, uses the high-frequency energy proportion of the current and the synchronous coupling of the current and the speed change to construct a multi-dimensional environmental state confidence degree, and aims to accurately identify the low-temperature oil pollution working condition and the obstacle to avoid frequent false alarms.

[0043] In one embodiment, the proportion of the high-frequency oscillation energy of the current data sequence is determined based on the following manner: The sum of squares of all values in the current data sequence is taken as the total energy of the current data sequence. The discrete wavelet transform technology (such as the multi-Besse wavelet transform) is used to perform multi-scale decomposition on the current data sequence to obtain the high-frequency detail coefficient sequence under each scale decomposition. The sum of squares of all values in the high-frequency detail coefficient sequence under the minimum scale decomposition is taken as the high-frequency oscillation energy. The ratio of the high-frequency oscillation energy to the total energy is determined as the proportion of the high-frequency oscillation energy of the current data sequence. The proportion can accurately quantify the micro-oscillation intensity in the motor running process and effectively reflect the non-stationary fluctuation caused by oil pollution adhesion.

[0044] For example, the current data sequence has a total energy of 30.43. The multi-scale decomposition is performed on the current data sequence by using the discrete wavelet transform technology, and the high-frequency detail coefficient sequence under the minimum scale decomposition is [0.1, -0.1, 0.1, -0.1, 0.05]. Therefore, the high-frequency oscillation energy is 0.0425, and the proportion of the high-frequency oscillation energy is 0.14%.

[0045] In one embodiment, the change trend correlation between the current data sequence and the speed data sequence is determined based on the following manner: First-order differencing operations are performed on the current and velocity data sequences respectively. That is, in each sequence, each value is subtracted from the previous value, resulting in two first-order differencing sequences. These sequences characterize the instantaneous rate of change and micro-fluctuation trend of the current and velocity data. The Pearson correlation coefficient between the two first-order differencing sequences is used as the correlation between the changing trends of the current and velocity data sequences. The Pearson correlation coefficient is a statistical indicator that measures the degree of linear correlation between two variables. Its value ranges from -1 to 1, and it is used to quantify the synchronicity and closeness of the changing trends of the current and velocity first-order differencing sequences.

[0046] In one embodiment, the confidence level of the environmental state from the start of the closing process to the end of the closing process is determined based on the following:

[0047] in, For the first Confidence of environmental states during the initial closure process to the end of closure. For normalization operations, For the first The degree of consistency of resistance drift during the entire closing process from the start to the end of the closing process. For the first High-frequency oscillation energy of the current data sequence from the start of the closing process to the end of the closing process. For the first The total energy of the current data sequence from the start of the closing process to the end of the closing process. For the first The correlation between the changing trends of the current data sequence and the velocity data sequence during the initial closing process to the final closing process. It is the absolute value symbol.

[0048] in, This provides a basic basis for judging the uniformity of macroscopic resistance. The larger the value, the more likely it is due to environmental interference, and the higher the energy ratio term. This accurately captures the microscopic high-frequency noise caused by oil viscosity in the current signal. The larger this ratio, the more the resistance source conforms to the characteristics of non-rigid fluid viscosity and low-temperature oil sludge, and the higher the confidence level of the corresponding environmental state. The correlation coefficient term... The coupling synchronization between current fluctuations and velocity fluctuations was verified from a dynamic perspective. The larger the value, the higher the correlation between the two, indicating that this is an inherent characteristic of the internal mechanical transmission of the system, rather than a collision with external discrete obstacles.

[0049] Determining the environmental state confidence level through multiplication means that the environmental state confidence level is only valid when the current data sequence simultaneously satisfies macroscopic drift consistency, microscopic viscous oscillations, and conforms to mechanical coupling laws. The larger the number of devices, the better. This ensures that the system only attributes the problem to environmental factors when it is fully confirmed that the issue is due to the equipment itself being old or dirty. This ensures that even in complex unmanned vending scenarios, the system firmly maintains its safety standards and avoids missing any actual incidents of people being trapped. Conversely, when... The smaller the value, the more likely it is an actual accident involving people or goods being trapped, rather than an accident caused by low-temperature oil contamination conditions.

[0050] S4: Based on the environmental state confidence of each closure process from start to stop, calculate the adaptive inertia weight of the PSO algorithm in each iteration.

[0051] In the anti-pinch control of smart cabinets, there is a balance between safety and usability. Excessive sensitivity can cause the door to fail to close at low temperatures, affecting food safety, while excessive insensitivity may injure users. Traditional fixed-parameter algorithms cannot simultaneously address both issues within a millisecond-level control cycle. If the system has confirmed with high confidence that the current increase in resistance is due to low temperature or oil contamination (environmental factors), the optimal strategy should be to quickly adjust the boundaries of the SVM classification model to encompass the environment, rather than performing a blind global search.

[0052] Therefore, this step establishes a dynamic switching mechanism for particle swarm search strategy based on environmental perception, which directly maps the calculated environmental state confidence to the inertial weights of the optimization algorithm, thereby realizing intelligent switching of the control strategy between steady-state adaptation and sudden high response.

[0053] The higher the confidence level of the environmental state, the greater the likelihood that the resistance encountered by the gate is caused by environmental factors. In this case, the PSO algorithm needs to reduce the inertia weight of the particles, weakening their ability to maintain their original motion state, and prompting the particle swarm to converge quickly to establish a new normal baseline. Conversely, if the confidence level of the environmental state is low, it indicates that there are more likely potential obstacles to the resistance. A larger inertia weight needs to be maintained for global exploration to avoid getting trapped in erroneous local optima. Based on the above logic, the adaptive inertia weight of each particle in each iteration of the particle swarm optimization algorithm is calculated.

[0054] In one embodiment, the adaptive inertia weights of the PSO algorithm in each iteration satisfy the following relationship:

[0055] in, In order to target the The PSO algorithm performs the parameter optimization process for the SVM classification model from the start of closure to the stop of closure. adaptive inertia weight of all particles in the next iteration, and are the maximum and minimum inertia weight of PSO algorithm respectively, is the environmental state confidence of the start of the closing process to the stop of the closing process.

[0056] The relationship establishes a negative feedback regulation mechanism between environmental certainty and the exploration ability of PSO algorithm. When approaches 1, it indicates that the system is highly confident that the current high resistance is a normal phenomenon caused by low temperature adhesive tape or rail oil stains, at this time is compressed to near the minimum inertia weight, which suppresses the global flight inertia of the particles at the algorithm level, forcing the particle swarm to conduct high-density local fine mining near the current solution space, so as to find a group of SVM parameters that can contain the current high resistance in a very short time, ensuring that the cabinet door can overcome the environmental resistance and close smoothly; on the contrary, when is lower, approaching 0, it means that it is more likely to exist sudden foreign matter or unknown situation, and it is more likely to be a situation of pinching or goods, maintains a high value, approaching the maximum inertia weight, giving the particles a strong global traversal ability, preventing the model from converging to the wrong classification boundary too early, and ensuring the highest sensitivity to potential pinching or goods risk. This dynamic adjustment mechanism improves the sensitivity of the anti-pinch control.

[0057] S5: parameter optimization of SVM classification model using adaptive inertia weight to optimize the SVM classification model.

[0058] Since the running environment of the intelligent vending cabinet is dynamically changing, such as day and night temperature difference, seasonal replacement, equipment aging, etc., which means that the definition of normal door closing is not fixed, and once the traditional offline training model is deployed, it is difficult to adapt to the new high resistance normal, which is the reason for the high false alarm rate. This step builds a dynamic training sample library, and combines the adaptive inertia weight calculated in the foregoing to drive the PSO algorithm to perform online optimization on the penalty factor and the kernel function parameter of the SVM classifier, the core purpose of which is to use the recently occurred high resistance data judged as environmental factors to train the SVM model, so that its classification boundary can dynamically contain environmental changes, while using historical abnormal data to maintain the safety bottom line.

[0059] In one embodiment, the method for parameter optimization of the SVM classification model using the adaptive inertia weight is: The training sample library is constructed, including: current data sequences and speed data sequences of a plurality of start closing to stop closing processes of the intelligent cabinet in a factory standard state, current data sequences and speed data sequences of a plurality of start closing to stop closing processes under a pinch abnormality, and current data sequences and speed data sequences of a plurality of processes from start closing to stop closing, in which the cabinet door is in a normal closing state, and the resistance change is judged to be caused by environmental factors in the last N times; A fitness function of a particle swarm optimization algorithm is defined, which represents cross-validation average accuracy of a support vector machine parameter combination represented by a particle in the training sample library, wherein each particle represents a set of support vector machine penalty factors and kernel function parameters; In each iteration, the particle velocity and position are updated, the motion state of the particle is updated by using the adaptive inertia weight, and the optimal penalty factor and the optimal kernel function parameter are searched in the preset parameter space; when the preset maximum iteration number is reached, the optimal penalty factor and the optimal kernel function parameter are obtained, and a SVM classification model is constructed based on the optimal penalty factor and the optimal kernel function parameter, so as to realize optimization of the SVM classification model.

[0060] A specific example is as follows: First, initialization is set: In order to meet the real-time requirement of the embedded system under limited computing power, the population size of the PSO algorithm is set to 30, the maximum iteration number is set to 50, the learning factor is set to 1.5, in order to avoid divergence of parameter search, the search space of the penalty factor of the SVM classification model is limited to , and the search space of the kernel function parameter of the SVM classification model is ; Next, a sample library is constructed: the sample library consists of a static baseline and a dynamic component. The static baseline contains 200 current and speed data sequences of the smart cabinet's start-to-stop closing process under factory standard conditions. Each process's current and speed data sequence constitutes a sample and is marked as a normal sample. The static baseline also includes 200 current and speed data sequences of the smart cabinet's start-to-stop closing process (historical data) when the cabinet is in a clamping or cargo-related state. Again, each process's current and speed data sequence constitutes a sample and is marked as an abnormal sample. All samples in the static baseline are used to lock the SVM classification. The model has basic anti-pinch capabilities; the dynamic part adopts a first-in-first-out sliding window mechanism with a capacity of 50. When the cabinet door is closed from the start to the stop of the current closure process, it only stores the current and velocity data sequences of the 50 most recent processes that the system judges as having resistance changes due to environmental factors and where the cabinet door is in a normal closed state. Similarly, the current and velocity data sequences of a process are treated as a sample and marked as a normal sample. The sliding window mechanism ensures that the model has a forgetting function. When the temperature rises or the oil is cleaned, the old high-resistance samples will be squeezed out by the new samples, and the model will automatically return to a high-sensitivity state.

[0061] Next, the fitness function is constructed: the fitness function serves as a bridge connecting the physical environment and mathematical optimization, and its calculation is directly derived from the training sample library constructed in the previous step. To ensure that the SVM model can accurately encompass the samples corresponding to the current low-temperature oil pollution environment, the fitness function... Defined as the combination of SVM parameters represented by the particle (penalty factor) and kernel function parameters The average cross-validation accuracy on the training set, and the fitness function. Defined as the average cross-validation accuracy of the SVM parameter combination represented by the particles on a training set containing recent measured high drag data:

[0062] in, This represents the fitness value of a set of SVM parameters corresponding to a particle in the PSO algorithm. The larger the value, the better the classification performance of that set of parameters. This indicates the number of folds in cross-validation (averaging the sample library into folds). share), This is used to evaluate the model's generalization ability and avoid overfitting. Indicates the first The total number of samples in the validation set during cross-validation includes normal samples from the factory baseline, abnormal samples from historical personnel or goods, and the most recent normal samples affected only by dynamic environments. Indicates the first The number of samples misclassified by the SVM model in the first cross-validation, i.e. the number of samples affected by the low-temperature oil stain misjudged as abnormal or the number of real samples misjudged as normal.

[0063] Since the recently occurred normal samples affected by the low temperature are added to the training sample library and are forcibly marked as normal samples, if the SVM parameter combination (penalty factor and kernel function parameter ) represented by the particle still follows the old standard, it will misjudge these normal samples with high resistance as abnormal, which will cause the number of misclassified samples in the formula to increase sharply, thereby causing the fitness value to decrease greatly. In order to pursue a higher value, the PSO algorithm will be forced to eliminate these old parameters and actively search for new parameters capable of containing these high-resistance samples. Therefore, the composition of the sample library (especially the addition of the dynamic part) determines the evaluation direction of the fitness function, thereby driving the algorithm to achieve self-adaptation to the current environment.

[0064] The physical meaning of the fitness function is that only when a certain parameter combination can correctly identify both the abnormal sample data of the historical inclusions / cargos contained in the static reference (prevent inclusions or cargos) and the normal sample data affected only by environmental interference contained in the dynamic part (prevent misjudgment of environmental impact as inclusions or cargos), the value of the fitness function will be the highest, thereby the parameter combination will be selected as the global optimal position of the PSO algorithm and become the gravitational center attracting all particles.

[0065] Subsequently, adaptive updating is performed: in the process of iterative search of the PSO algorithm for the optimal parameter combination of the SVM classification model, in each iteration, first, the global optimal position of the PSO algorithm is updated according to the above fitness function, and then the adaptive inertia weight calculated in step S4 is substituted when updating the particle speed. When the environmental confidence is high, the adaptive inertia weight automatically becomes small, weakening the flight inertia of the particle, so that it performs a small local disturbance near the global optimal position, achieving rapid convergence; when the environmental confidence is low, the adaptive inertia weight maintains a larger value, maintaining the global traversal ability of the particle, avoiding falling into a local optimum. It should be noted that for the qth start-closing to stop-closing process, in the process of parameter optimization of the PSO algorithm for the SVM classification model, the adaptive inertia weight of all particles is the value calculated in step S4, regardless of whether it is the 1st iteration or the 50th iteration. The adaptive inertia weight is only related to the environmental diagnosis result, and does not decay with the iteration number. This is the difference between the present application and the traditional PSO algorithm, and is also the core of environmental self-adaptation. ​

[0066] Finally, when the number of iterations reaches 50 and the output finally converges, the penalty factor and kernel function parameters in the parameter combination corresponding to the global optimal position of the PSO algorithm are used as the optimal penalty factor and optimal kernel function parameters, respectively. The final SVM classification model is constructed using the optimal penalty factor and optimal kernel function parameters for real-time anti-pinch control in step S6. S6: Using the output of the optimized SVM classification model, the state of the cabinet door from the start of closing to the stop of closing is determined as normal or abnormal, and anti-pinch control is performed on the smart cabinet door based on the state.

[0067] In one embodiment, anti-pinch control of the smart cabinet door based on the state includes: The current and speed data sequences from the start to the stop of each cabinet door closure process are input into the optimized SVM classification model. If the SVM classification model outputs an abnormal classification result, it is determined that there is a risk of trapping a person, the cabinet door is in an abnormal closing state, the motor is controlled to stop rotating and rotate in the reverse direction, causing the cabinet door to spring back, minimizing the risk of personal injury. If the SVM classification model outputs a normal classification result, it is determined that the resistance change is due to environmental factors, the current increase in resistance is due to environmental drift, the cabinet door is in a normal closing state. At this time, the main control unit will ignore the conventional current threshold protection and can selectively and appropriately increase the upper limit of the motor's output torque to overcome the hardening resistance of the low-temperature rubber strip, forcibly driving the cabinet door to continue to perform the closing action until the locking signal is triggered.

[0068] like Figure 2 As shown, the horizontal axis represents time, and the vertical axis represents the output score of the SVM classification model on the cabinet door's operating status. In other words, the operating status of the cabinet door is evaluated based on the output classification results of the SVM classification model. The higher the output score, the greater the risk of trapping people or goods. Existing technologies are affected by environmental resistance and fluctuate around the anti-pinch trigger threshold, which is prone to false alarms. This invention uses environmental adaptation to lower the score and eliminate false alarms. The anti-pinch trigger threshold is only exceeded when a person is actually trapped. While preventing false alarms, it ensures accurate triggering of real trapping accidents.

[0069] This invention also provides an intelligent cabinet door anti-pinch control system based on multi-source sensor fusion. The intelligent cabinet door anti-pinch control system includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, they implement all or part of the steps of the intelligent cabinet door anti-pinch control method. The processor can be a central processing unit (CPU), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), or a field-programmable gate array (FPGA). The processor is preferably a GPU to achieve parallel acceleration. The memory can include volatile memory (such as RAM) and non-volatile memory (such as ROM or flash memory).

[0070] The above description is only the preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A multi-source sensor fusion-based intelligent cabinet door anti-pinch control method, characterized in that, The method comprises the following steps: obtaining a current data sequence and a speed data sequence of the smart cabinet door in each start-to-stop closing process, and a reference current data sequence composed of the average of current data of multiple start-to-stop closing processes of the smart cabinet door in a factory standard state; determining the consistency degree of resistance drift in each start-to-stop closing process based on the numerical difference between the current data sequence and the reference current data sequence; extracting the proportion of high-frequency oscillation energy of the current data sequence, calculating the trend correlation between the current data sequence and the speed data sequence, and comprehensively determining the environmental state confidence of each start-to-stop closing process based on the consistency degree of resistance drift, the trend correlation, and the proportion of high-frequency oscillation energy; based on the environmental state confidence, calculating the adaptive inertia weight of the PSO algorithm at each iteration, optimizing the SVM classification model by parameter optimization of the SVM classification model using the adaptive inertia weight, and using the output of the optimized SVM classification model to determine the state of each start-to-stop closing process of the cabinet door as normal or abnormal, and performing anti-pinch control on the smart cabinet door according to the state.

2. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The degree of uniformity of drag drift is determined based on the following method: for the first Alignment is performed on the current and velocity data sequences from the start of closure to the end of closure; drag drift consistency is calculated. ;in, For the first The degree of consistency of resistance drift during the entire closing process from the start to the end of the closing process. For the first The length of the current data sequence from the start of the closing process to the end of the closing process. For the first The current data sequence from the start of closing the secondary cabinet door to the stop of closing. A number, The first in the reference current data sequence A number, For the first The overall offset between the current data sequence from the start of closing of the secondary cabinet door to the stop of closing and the reference current data sequence is the value of the second current data sequence. The current data sequence from the start of the closing process to the stop of the closing process is subtracted element by element from the reference current data sequence, and then the average is calculated. To prevent constants with a denominator of zero, This is a normalization operation.

3. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The method for extracting the proportion of high-frequency oscillation energy of the current data sequence is as follows: the square sum of all values in the current data sequence is taken as the total energy of the current data sequence; the discrete wavelet transform technology is used to perform multi-scale decomposition on the current data sequence to obtain high-frequency detail coefficient sequences at each scale decomposition; and the square sum of all values in the high-frequency detail coefficient sequence at the smallest scale decomposition is taken as the high-frequency oscillation energy. The ratio of the high-frequency oscillation energy to the total energy is determined as the proportion of high-frequency oscillation energy of the current data sequence.

4. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The trend correlation between the current data sequence and the speed data sequence is determined based on the following method: performing first-order difference operation on the current data sequence and the speed data sequence respectively to obtain two first-order difference sequences, and taking the Pearson correlation coefficient between the two first-order difference sequences as the trend correlation between the current data sequence and the speed data sequence.

5. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The environmental state confidence of each start-to-stop closing process satisfies the following relationship: ; wherein, is the environmental state confidence of the is the normalization operation, is the resistance drift consistency of the is the high frequency oscillation energy of the current data sequence of the is the total energy of the current data sequence of the is the trend correlation between the current data sequence and the speed data sequence of the is the absolute value sign.

6. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The adaptive inertia weight of the PSO algorithm at each iteration satisfies the following relationship: ;in, In order to target the The PSO algorithm performs the parameter optimization process for the SVM classification model from the start of closure to the stop of closure. The adaptive inertia weights of all particles in the next iteration. and These are the maximum and minimum inertia weights preset for the PSO algorithm, respectively. For the first Confidence of environmental states during the process from the start of closure to the end of closure.

7. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The method for parameter optimization of the SVM classification model using the adaptive inertia weight is as follows: The training sample library is constructed, including: a plurality of normal sample data in a factory standard state, a plurality of abnormal sample data when the cabinet door generates a pinch / goods, and a plurality of normal sample data recently determined as a resistance change source from environmental factors, and the cabinet door is in a normal closed state, wherein the normal / abnormal sample data in the training sample library each contains a current data sequence and a speed data sequence of a cabinet door starting to close to stopping closing process; the fitness function of the PSO algorithm is defined, which represents the cross-validation average accuracy of the parameter combination represented by the particle on the training sample library; the particle state is updated by using the adaptive inertia weight in the iterative optimization process of the PSO algorithm to search for the optimal parameter combination, and when the maximum iteration number is reached, the SVM classification model is constructed by using the optimal parameter combination.

8. The intelligent cabinet door anti-pinch control method of claim 2, wherein, The alignment operation is performed based on the following manner: The speed data sequence is used for discrete time integration to obtain the cumulative displacement amount of the door body at each time of the starting closing to stopping closing process, so as to construct the corresponding relationship of the cumulative displacement amount of the door body, the current and the speed at each time; a standardized door body stroke space coordinate system is set, and the door body stroke space coordinate system is divided into a plurality of equally spaced standard space sampling points; the current and the speed corresponding to each standard space sampling point are calculated by using the cumulative displacement amount of the door body as an index and applying a linear interpolation algorithm, so that the speed data sequence and the current data sequence one-to-one correspond based on the standard space sampling points, so as to realize the alignment operation.

9. The intelligent cabinet door anti-pinch control method of claim 1, wherein, The state of the starting closing to stopping closing process of each cabinet door is normal or abnormal by using the output of the optimized SVM classification model, and the intelligent cabinet door is prevented from being pinched according to the state, including: The current data sequence and the speed data sequence of the starting closing to stopping closing process of each cabinet door are input into the optimized SVM classification model; if the classification result output by the SVM classification model is abnormal, it is determined that there is a pinching risk, the cabinet door is in an abnormal closing state, the motor is controlled to stop rotating and reverse rotation, and the cabinet door is bounced back; if the classification result output by the SVM classification model is normal, it is determined that the resistance change is caused by environmental factors, and the cabinet door is in a normal closing state.

10. An intelligent cabinet door anti-pinch control system based on multi-source sensing fusion, characterized in that, The intelligent cabinet door anti-pinch control system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the steps of the intelligent cabinet door anti-pinch control method in any one of claims 1-9.

Images