Refrigerator door body vibration induction detection control system

Abstract

The invention discloses a refrigerator door body vibration induction detection control system, and relates to the technical field of refrigerator detection, the system comprises the following components: a vibration data acquisition subsystem, a data processing and analysis subsystem, a machine learning analysis module, a fault early warning and maintenance prompt subsystem and a data storage and management subsystem; through the high-precision piezoelectric vibration sensor and an advanced signal processing algorithm, the vibration state of the refrigerator door body can be monitored in real time, the system can accurately capture the key information of the vibration amplitude, frequency, phase and waveform of the door body, and the vibration data is deeply analyzed by using a machine learning algorithm, so that the accuracy of the vibration state of the refrigerator door body is improved. The door body fault early warning and accurate positioning function is achieved, the problems that due to the door body fault, the sealing performance of the refrigerator is reduced, and energy consumption is increased are effectively solved, the service life of the refrigerator is prolonged, and meanwhile the maintenance cost and time cost of a user are reduced.

Description

Technical Field

[0001] The invention relates to the technical field of refrigerator detection, and in particular to a refrigerator door vibration induction detection control system. Background Art

[0002] With the popularization of modern household appliances and the development of intelligent technology, refrigerators, as indispensable electrical equipment in family life, are increasingly concerned by consumers for their performance and reliability. As an important part of the refrigerator, the refrigerator door is not only related to the overall appearance of the refrigerator, but also directly affects the sealing performance and energy consumption of the refrigerator. However, in actual use, the refrigerator door often vibrates abnormally due to long-term use, material aging, improper installation or external force impact, which leads to problems such as poor sealing, increased noise and increased energy consumption. In severe cases, it may even cause damage to the door, affecting the normal use of the refrigerator.

[0003] Traditional technology has its shortcomings. First, the sensors usually used in traditional technology have limited accuracy, making it difficult to accurately capture subtle changes in the vibration of the refrigerator door, resulting in low accuracy of fault warning. Second, the data analysis methods of traditional technology are relatively simple and lack the support of deep learning and machine learning. They are unable to effectively mine and analyze complex vibration data, thus limiting the accuracy of fault prediction.

[0004] In summary, the existing refrigerator door detection technology has the problems of low detection efficiency, low fault warning accuracy and single maintenance reminder method, which cannot meet the high requirements of modern families for refrigerator performance and stability. Therefore, it is particularly important to develop a refrigerator door vibration sensing detection and control system. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and to provide a refrigerator door vibration sensing detection and control system, which can realize real-time monitoring of the vibration state of the refrigerator door and fault warning through high-precision piezoelectric vibration sensors, advanced signal processing algorithms and machine learning technologies, thereby improving the stability and service life of the refrigerator and reducing the user's maintenance costs and time costs.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a refrigerator door vibration sensing detection and control system, the system comprising the following components: a vibration data acquisition subsystem, a data processing and analysis subsystem, a machine learning analysis module, a fault warning and maintenance prompt subsystem, and a data storage and management subsystem;

[0007] The vibration data acquisition subsystem: This subsystem includes a plurality of highly sensitive piezoelectric vibration sensors. The sensors are precisely arranged on the refrigerator door body according to the mechanical structure characteristics of the refrigerator door body and the vibration modal analysis results. The specific positions include the four corners of the door body, the door axis connection, the center area of ​​the door body, the three-division points along the height direction of the door body, and the two-division points along the width direction of the door body. The sensors collect the vibration data of the refrigerator door body in a periodic sampling manner. The sampling frequency f s According to the natural frequency range f0 of the door body, f s =10×f0, the vibration data collected covers the vibration amplitude A, frequency F, phase and vibration waveform w(t) information, and transmit the collected data to the data processing and analysis subsystem;

[0008] The data processing and analysis subsystem includes:

[0009] The raw vibration data received from the vibration data acquisition subsystem is processed in the following steps: A hybrid domain noise suppression algorithm is used, which combines the advantages of the time domain and the frequency domain. The formula is: Where y(t) is the processed signal, x(t) is the original signal, and h i (t) is the impulse response of the i-th noise filter in the time domain, α i is the corresponding weighting coefficient, α i The method of determining is: in is the cross power spectral density of the original signal x(t) and the i-th noise source in the frequency domain, S x (f) is the power spectrum density of the original signal. The algorithm first performs Fourier transform on the original signal to convert it into the frequency domain. By analyzing the power spectrum characteristics of different frequency bands, the noise frequency band is identified and the corresponding filter h is designed. i (t) and its weighting coefficient α i , to suppress various noise sources;

[0010] Using the feature scaling normalization method, for each vibration data dimension d, the normalization formula is: where x d is the original data of a certain dimension, x dj is the jth data point of the dimension, M is the total number of data points of the dimension. This method uses logarithmic transformation to map data of different magnitudes to a relatively compact range, and at the same time considers the overall distribution of data points, so that the data can better reflect their relative relationship after normalization. The data processed as above is divided into data segments according to the time series based on events, that is, according to the alternating cycles of the opening and closing events of the refrigerator door and the static state, the data segments are divided into standard data samples;

[0011] The machine learning analysis module adopts an analysis algorithm based on evolutionary fuzzy neural network. The fuzzy rule generation process of the algorithm is as follows: the normalized data samples are taken as input, each data sample is regarded as a potential fuzzy set, and the initial fuzzy rule set is generated using a genetic algorithm, where each fuzzy rule R k The form is: IF A k ANDB k AND C k AND D k THEN E k , A k , B k , C k , D k Represents amplitude A, frequency F, phase and the fuzzy set of waveform w(t),

[0012] E k represents the fault category fuzzy set, and the membership function of the fuzzy set adopts the Gaussian function: in and The mean and standard deviation are determined by clustering analysis of a large amount of historical normal data. The fuzzy rule set is optimized using an evolutionary algorithm. The optimization objective function is: where y i is the expected output, is the predicted output of the network, K is the number of fuzzy rules L(R k ) is the fuzzy rule R k The complexity penalty term, β is the complexity penalty coefficient, and its value is adjusted through experiments on historical data of different refrigerator brands and models. By continuously evolving the fuzzy rule set, it can accurately learn the vibration mode and characteristics of the door body during normal operation. After receiving the real-time data sample, the data is pattern recognized and analyzed, and the difference between the real-time data and the normal model is calculated to determine whether the current operating state of the door body is normal. Once an abnormality is found, the abnormal data is deeply mined to predict the possible fault type of the door body and the specific time point when the fault may occur;

[0013] The fault warning and maintenance prompt subsystem: based on the prediction results of the machine learning analysis module, when it is determined that the door body may fail within a specific time period in the future, detailed maintenance prompt information is sent to the user through the visual interface on the refrigerator operation panel, the mobile phone application bound to the refrigerator, and voice prompts;

[0014] The data storage and management subsystem adopts a hybrid storage architecture, including local non-volatile memory and cloud storage services, to store historical vibration data collected by the vibration data acquisition subsystem, preprocessed data samples, model parameters trained by the machine learning analysis module, and fault prediction information generated by the fault warning and maintenance prompt subsystem, and has data cleaning, optimization and encrypted storage functions, as well as local and cloud data synchronization functions.

[0015] Furthermore, the piezoelectric material of the piezoelectric vibration sensor in the vibration data acquisition subsystem is selected as a new type of lead lanthanum zirconate titanate ceramic material, which has a higher piezoelectric constant and a lower mechanical quality factor, and can convert weak mechanical vibrations into stronger electrical signals. The packaging structure of the sensor is specially designed and adopts double-layer protection. The outer layer is a metal shielding cover, which can effectively prevent electromagnetic interference, and the inner layer is a flexible shock-absorbing material to avoid the influence of external mechanical shock on the sensor. At the same time, the sensor is connected to the door body with an elastic adhesive, and its elastic modulus is matched according to the elastic modulus of the door body material to ensure good coupling between the sensor and the door body, which can truly reflect the vibration state of the door body. In addition, the installation angle of the sensor is fine-tuned according to the main direction of the door body vibration and the modal analysis results to maximize the reception of vibration energy and improve the collection efficiency of vibration signals.

[0016] Furthermore, the mixed domain noise suppression algorithm in the data preprocessing module of the data processing and analysis subsystem determines the noise filter h i (t), it is dynamically adjusted according to the different operating modes of the refrigerator. In the refrigeration mode, the operation of the compressor will generate interference of specific frequencies. Through real-time monitoring of the compressor operating frequency, it is taken as the key noise frequency band and a targeted notch filter is designed as h i (t), whose center frequency is the same as the compressor frequency, and the bandwidth is determined according to the half-power bandwidth of the compressor vibration spectrum. For environmental noise, according to the data of the environmental sensor, when the ambient temperature and noise level change, the corresponding filter parameters are automatically adjusted to adapt to the noise characteristics in different environments. At the same time, when calculating the weighting coefficient α i When the door is opened and closed, the impact of the number of times the door is opened and closed on the noise is taken into consideration. Every time the door is opened and closed, the noise spectrum is updated in real time according to the speed and force of opening and closing, so as to adjust the weighting coefficient more accurately and improve the pertinence and effectiveness of noise suppression.

[0017] Furthermore, the feature scaling normalization method in the data preprocessing module of the data processing and analysis subsystem assigns different weights to data of different dimensions when calculating normalized data. By analyzing the fault sensitivity of historical data, the weight of amplitude A is determined to be w A =0.4, the weight of frequency F is w F=0.3, phase The weight of The weight of the waveform w(t) is w w =0.1. This weight distribution is based on the variance contribution rate and correlation analysis of each dimension data under different failure modes. The amplitude shows a higher variance in the door body seal failure failure, so it is given a higher weight. In addition, for the normalized data, data smoothing is further adopted, using a moving average filter: Where M is the smoothing window size, which is determined according to the sampling frequency of the data and the main frequency components of the door vibration to eliminate high-frequency jitter in the data and improve the stability of the data for subsequent analysis.

[0018] Furthermore, the number of input layer nodes of the evolutionary fuzzy neural network in the machine learning analysis module of the data processing and analysis subsystem is determined according to the data dimension and the number of data points in the data segment. For each data segment, it is divided into multiple sub-intervals, each sub-interval corresponds to an input node, and the number of input nodes n=D×S, where D is the data dimension (amplitude A, frequency F, phase φ, waveform w(t)), S is the number of sub-intervals, and the sub-intervals are divided based on the typical cycle of door body vibration and the statistical distribution characteristics of historical data. In the process of generating fuzzy rule sets by genetic algorithm, the size of the initial population is determined according to the complexity of the refrigerator door body and the diversity of historical fault data. For complex door body structures and diverse fault data, the initial population size is set to a larger value to ensure the diversity of the rule set. In the evolutionary process, an adaptive crossover probability P is used. c and mutation probability P m , P c and P m Adjust according to the fitness variance of the current population. When the fitness variance is small, increase P c and P m , in order to jump out of the local optimal solution and ensure that the evolution of the rule set can move towards the global optimal direction.

[0019] Furthermore, the visualization interface of the fault warning and maintenance prompt subsystem adopts an ultra-high-definition display screen on the refrigerator operation panel, which has a touch interaction function. The interface layout is divided into multiple modules, including a real-time vibration data monitoring module, a fault prediction details module, a maintenance operation guidance module and a user feedback module. The real-time vibration data monitoring module displays various indicators of the current vibration data in the form of dynamic graphics and real-time data lists, and can display the data change trend in the past hour. The fault prediction details module displays the fault risk distribution at different time points in the future in the form of a timeline and a heat map. The darker the color, the higher the fault risk. The maintenance operation guidance module provides detailed illustrated maintenance steps according to the predicted fault type, including the required tools, operation sequence and precautions. For parts that need to be replaced, a three-dimensional model of the parts and an assembly animation demonstration are provided. The user feedback module allows users to evaluate and provide feedback on maintenance prompts. The system will collect this feedback information for further optimizing the fault prediction model and maintenance prompt information.

[0020] Furthermore, when the fault warning and maintenance reminder subsystem sends information to the mobile phone application, it will also send augmented reality guidance information. After the user opens the APP, he can aim the mobile phone camera at the refrigerator door and see the virtual fault mark and maintenance operation instructions superimposed on the door. These instructions are accurately displayed according to the predicted fault type and location. The APP also has a remote diagnosis function. When the user turns on this code, the current vibration data and system status information will be sent to the remote server, and professional technicians will provide users with more accurate maintenance suggestions through remote analysis. In addition, the APP supports multi-language services and can automatically switch the display language according to the system language settings of the user's mobile phone, which is convenient for users of different languages. At the same time, it provides user-defined notification settings, allowing users to set reminder times and reminder methods for different fault types to meet personalized needs.

[0021] Furthermore, the local non-volatile memory of the data storage and management subsystem adopts a three-level storage architecture, which is divided according to the importance and access frequency of the data. The first level stores the latest high-priority data, vibration data within the past week and current fault prediction information, and adopts a high-speed cache mechanism to improve the reading and writing speed of the data. The second level stores moderately important data in the past month, and adopts redundant storage technology to ensure data reliability. The third level stores historical data and adopts compression storage technology to reduce storage space. For cloud storage services, a distributed storage cluster is used. Each storage node allocates storage locations according to the hash value and timestamp of the data, and a consistent hashing algorithm is used to achieve balanced distribution of data. Data encryption adopts multi-key encryption technology, and different encryption keys are used for different types of data. The generation and management of keys are performed through a hardware security module to ensure data security and manageability.

[0022] Furthermore, when the data processing and analysis subsystem conducts in-depth mining of abnormal data to predict the fault type and time point, it also adopts an auxiliary judgment mechanism based on multi-sensor fusion, combines the data of the temperature sensor, humidity sensor and pressure sensor installed inside the refrigerator, and establishes a multi-sensor data fusion model. After normalization and feature extraction of the data of each sensor, the Bayesian reasoning method is used for fusion, and the fusion formula is: Where F represents the fault type, D represents the data after multi-sensor fusion, P(F) is the prior probability of the fault type, and P(D|F) is the probability of data D appearing when fault F occurs. Through multi-sensor data fusion, the door status can be evaluated from different angles, supplementing the judgment results of the evolved fuzzy neural network and improving the comprehensiveness and accuracy of fault prediction.

[0023] Compared with the prior art, this refrigerator door vibration sensing detection control system has the following beneficial effects:

[0024] 1. The present invention can monitor the vibration state of the refrigerator door in real time through a high-precision piezoelectric vibration sensor and an advanced signal processing algorithm. The system can accurately capture the amplitude, frequency, phase and waveform key information of the door vibration, and use a machine learning algorithm to conduct an in-depth analysis of the vibration data to achieve early warning and precise positioning of door failures. This function effectively avoids the problems of reduced sealing performance and increased energy consumption of the refrigerator due to door failures, thereby extending the service life of the refrigerator and reducing the user's maintenance and time costs.

[0025] 2. Through the visual interface on the refrigerator operation panel, the mobile phone application bound to the refrigerator, and voice prompts, the system can send detailed maintenance prompt information to the user in a timely manner, including fault prediction details and maintenance operation instructions. In particular, in the mobile phone application, the system also provides augmented reality guidance information and remote diagnosis functions, so that users can understand the fault conditions of the refrigerator door more conveniently and intuitively, and obtain professional maintenance suggestions. This function not only improves the user experience, but also enhances the practicality and intelligence level of the system.

[0026] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 This is a flow chart of a refrigerator door vibration sensing detection control system;

[0029] Figure 2 This is the process operation diagram of the machine learning analysis module and the fault warning and maintenance prompt subsystem. DETAILED DESCRIPTION

[0030] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0031] Embodiment 1

[0032] This embodiment describes that a user in an ordinary household frequently opens and closes a refrigerator door.

[0033] The vibration data acquisition subsystem in the refrigerator door vibration sensing detection and control system has multiple highly sensitive piezoelectric vibration sensors that are precisely arranged at the four corners of the door and the door shaft connection based on the mechanical structure characteristics of the door and the results of vibration modal analysis. The piezoelectric material of the sensor is a new type of lead lanthanum zirconate titanate ceramic material, with a double-layer protective packaging structure, and is connected to the door body through an elastic adhesive. The installation angle is fine-tuned, and the sampling frequency f s =10×f0, the vibration data collected covers the vibration amplitude A, frequency F, phase And the vibration waveform w(t) information, and the collected data is transmitted to the data processing and analysis subsystem. After the data processing and analysis subsystem receives the original vibration data, it first uses the mixed domain noise suppression algorithm to process the noise. The algorithm formula is in

[0034] In the cooling mode, the system detects that the compressor operates at a frequency of 50 Hz. Taking this as the key noise frequency band, a notch filter with a center frequency of 50 Hz and a bandwidth determined according to the half-power bandwidth of the compressor vibration spectrum is designed as h i (t), and dynamically adjust the filter parameters and weighting coefficients according to the environmental sensor data and the door opening and closing conditions. Then, the feature scaling normalization method is used to normalize the amplitude A, frequency F, and phase The four dimensions of data, w(t), are assigned weights w A =0.4, w F =0.3, w w =0.1, the normalized formula is And use moving average filter The normalized data is smoothed to eliminate high-frequency jitter. The processed data is divided into data segments according to the refrigerator door opening and closing events and the static state alternating cycle to form a standard data sample, which is input into the machine learning analysis module. The machine learning analysis module adopts an analysis algorithm based on evolutionary fuzzy neural network. The number of input layer nodes n = D × S is determined according to the data dimension and the number of sub-intervals of the data segment. The genetic algorithm generates the initial fuzzy rule set. The fuzzy rule form is: IF A k AND B k AND C k AND D k THEN E k , A k , B k , C k , D k Represents amplitude A, frequency F, phase and the fuzzy set of waveform w(t), E k represents the fault category fuzzy set, and the membership function of the fuzzy set adopts the Gaussian function:

[0035] During the evolution process, the adaptive crossover probability P c and mutation probability P m According to the population fitness variance adjustment, when the user frequently opens and closes the refrigerator door, causing the door vibration data to be abnormal, the system determines that the door operation status is abnormal by calculating the difference between the real-time data and the normal model, and predicts that it may be a door seal failure, and serious problems may occur in the next 24 hours. After receiving the prediction results, the fault warning and maintenance prompt subsystem displays real-time vibration data monitoring, fault prediction details, maintenance operation instructions and user feedback module information through the ultra-high-definition touch screen of the refrigerator operation panel. At the same time, the mobile application sends augmented reality guidance information, and users can see virtual fault marks and maintenance operation instructions through the mobile phone camera. The APP also supports remote diagnosis and multi-language services. The data storage and management subsystem uses a hybrid storage architecture to store relevant data. The local non-volatile memory uses a three-level storage architecture, and the cloud uses a distributed storage cluster and performs data encryption.

[0036] Embodiment 2

[0037] This embodiment describes a large chain supermarket, where the fresh food area and the frozen food area are equipped with multiple large commercial refrigerators for storing various fresh foods, frozen ingredients and cold drinks. Due to the large customer flow and long business hours of the supermarket, the refrigerators need to be opened and closed frequently to meet the shopping needs of customers. At the same time, a large amount of goods of different weights are usually stored in the refrigerators, which makes the refrigerator door body bear greater pressure and complex vibration.

[0038] The freezer is equipped with a refrigerator door vibration sensing detection and control system. The piezoelectric vibration sensor in the vibration data acquisition subsystem uses a new type of lead lanthanum zirconate titanate ceramic material. Its packaging adopts a double-layer protection structure. The outer metal shielding cover effectively shields external electromagnetic interference, and the inner flexible shock-absorbing material buffers the impact of door vibration on the sensor. The sensor is installed on the door body through an elastic adhesive that matches the elastic modulus of the door material, and the installation angle is fine-tuned according to the main direction of door vibration and the modal analysis results. These sensors are precisely arranged at the four corners of the door body, the door axis connection, the center area of ​​the door body, the three-division point along the height direction of the door body, and the two-division point along the width direction of the door body. The sampling frequency is 10 times the natural frequency of the door body. The vibration data of the door body is collected in real time, covering the amplitude, frequency, phase and vibration waveform information, and the data is quickly transmitted to the data processing and analysis subsystem.

[0039] After receiving the raw vibration data, the data processing and analysis subsystem immediately starts processing. In response to the strong interference generated by the compressor during the refrigeration of the refrigerator, the system monitors the compressor operating frequency in real time, takes it as the key noise frequency band, and designs a targeted notch filter. For example, when the compressor operating frequency is 50Hz, the center frequency of the notch filter is also set to 50Hz, and the bandwidth is determined according to the half-power bandwidth of the compressor vibration spectrum, which effectively filters out the interference of this specific frequency. For environmental noise, the system automatically adjusts the corresponding filter parameters according to the data of the environmental sensor, such as the changes in ambient temperature and noise level. At the same time, considering the frequent opening and closing of the refrigerator door in the supermarket, the system updates the noise spectrum in real time according to the speed and strength of the opening and closing each time the door is opened and closed, and then accurately adjusts the weighting coefficient to more effectively remove various types of noise. When the data is normalized by feature scaling, it is processed according to the weights of amplitude, frequency, phase and waveform, and the normalized data is smoothed by a moving average filter to remove high-frequency jitter, and then the data segments are divided according to the refrigerator door opening and closing events.

[0040] The machine learning analysis module adopts an analysis algorithm based on evolutionary fuzzy neural networks to conduct in-depth learning of the processed data samples. The number of input layer nodes is determined according to the data dimension and the number of data points in the data segment. Each data segment is divided into multiple sub-intervals, and each sub-interval corresponds to an input node. In the process of generating fuzzy rule sets by genetic algorithms, in view of the relatively complex structure of commercial refrigerator doors and diverse fault conditions, the initial population size is set to a larger value to ensure the diversity of the rule set. During the evolution process, the adaptive crossover probability and mutation probability are adjusted according to the fitness variance of the current population. When the fitness variance is small, these two probabilities are increased to prompt the algorithm to jump out of the local optimal solution. After learning a large amount of data, the system successfully established a vibration model suitable for commercial refrigerator doors.

[0041] During the business operation of a certain day, the system detected abnormal vibration data of the door of a certain refrigerator. The machine learning analysis module conducted in-depth mining of the abnormal data, combined with the auxiliary judgment mechanism of multi-sensor fusion, integrated the data of the temperature sensor, humidity sensor and pressure sensor installed inside the refrigerator, and performed fusion analysis through the Bayesian reasoning method. It was finally predicted that the refrigerator door shaft would fail within three days due to severe wear and tear due to long-term frequent use, which might cause the door to fail to open and close normally.

[0042] The fault warning and maintenance reminder subsystem takes quick action. Through the ultra-high-definition touch screen on the refrigerator operation panel, the current vibration data indicators and the data change trend of the past hour are displayed in dynamic graphics and real-time data lists. In the fault prediction details module, the fault risk distribution at different time points in the future is presented in the timeline and heat map. The maintenance operation guidance module provides detailed graphic maintenance steps, required tools, operation sequence and precautions for door shaft failures, and displays the three-dimensional model and assembly animation demonstration of the door shaft components. At the same time, the mobile phone APP sends augmented reality guidance information to the supermarket maintenance personnel. After the maintenance personnel open the APP, they can point the mobile phone camera at the refrigerator door to see the virtual fault mark and maintenance operation instructions superimposed on the door body. The APP also has a remote diagnosis function. After the maintenance personnel turn on this function, the current vibration data and system status information will be sent to the remote server. Professional technicians will conduct remote analysis and provide more accurate maintenance suggestions for maintenance personnel. In addition, the APP supports multi-language services, and can automatically switch the display language according to the system language setting of the maintenance personnel's mobile phone. It also provides custom notification settings to facilitate maintenance personnel to set reminder times and methods for different types of faults.

[0043] After receiving the early warning information, the supermarket maintenance staff will arrange the maintenance work in time. After the maintenance is completed, the maintenance staff can evaluate and provide feedback on the maintenance process and results in the system's user feedback module. The data storage and management subsystem adopts a local three-level storage architecture and a cloud distributed cluster to store relevant data. The local first level stores the latest high-priority data, the second level uses redundant storage technology to store moderately important data for the past month, and the third level uses compression storage technology to store historical data. Cloud storage uses a consistent hashing algorithm to achieve balanced data distribution and uses multi-key encryption technology to ensure data security. The stored data will be used for subsequent optimization of the fault prediction model, further improving the system's fault detection and prediction capabilities in commercial refrigerator scenarios, ensuring the stable operation of supermarket refrigerators, and reducing commodity losses and operating losses caused by refrigerator failures.

[0044] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A refrigerator door vibration sensing detection control system, characterized in that: The system includes the following components: vibration data acquisition subsystem, data processing and analysis subsystem, machine learning analysis module and fault warning and maintenance prompt subsystem, as well as data storage and management subsystem; The vibration data acquisition subsystem: This subsystem includes a plurality of highly sensitive piezoelectric vibration sensors. The sensors are precisely arranged on the refrigerator door body according to the mechanical structure characteristics of the refrigerator door body and the vibration modal analysis results. The specific positions include the four corners of the door body, the door axis connection, the center area of ​​the door body, the three-point point along the height direction of the door body, and the two-point point along the width direction of the door body. The sensors collect the vibration data of the refrigerator door body in a periodic sampling manner. The sampling frequency f s According to the natural frequency range f0 of the door body, f s =10×f0, the vibration data collected covers the vibration amplitude A, frequency F, phase and vibration waveform w(t) information, and transmit the collected data to the data processing and analysis subsystem; The data processing and analysis subsystem includes: The raw vibration data received from the vibration data acquisition subsystem is processed. The specific steps are as follows: Use the hybrid domain noise suppression algorithm, which combines the advantages of time domain and frequency domain. The formula is: Where y(t) is the processed signal, x(t) is the original signal, and h i (t) is the impulse response of the i-th noise filter in the time domain, α i is the corresponding weighting coefficient, α i The method of determining is: in is the cross power spectral density of the original signal x(t) and the i-th noise source in the frequency domain, S x (f) is the power spectral density of the original signal. The algorithm first performs Fourier transform on the original signal to convert it into the frequency domain. By analyzing the power spectrum characteristics of different frequency bands, the noise frequency band is identified and the corresponding filter h is designed. i (t) and its weighting coefficient α i , to suppress various noise sources; Using the feature scaling normalization method, for each vibration data dimension d, the normalization formula is: where x d is the original data of a certain dimension, x dj is the jth data point of the dimension, M is the total number of data points of the dimension. This method uses logarithmic transformation to map data of different magnitudes to a relatively compact range, and at the same time considers the overall distribution of data points, so that the data can better reflect their relative relationship after normalization. The data processed as above is divided into data segments according to the time series based on events, that is, according to the alternating cycles of the opening and closing events of the refrigerator door and the static state, the data segments are divided into standard data samples; The machine learning analysis module adopts an analysis algorithm based on evolutionary fuzzy neural network. The fuzzy rule generation process of the algorithm is as follows: the normalized data samples are taken as input, each data sample is regarded as a potential fuzzy set, and the initial fuzzy rule set is generated using a genetic algorithm, where each fuzzy rule R k The form is: IF A k AND B k ANDC k AND D k THEN E k , A k , B k , C k , D k Represents amplitude A, frequency F, phase and the fuzzy set of waveform w(t), E k represents the fault category fuzzy set, and the membership function of the fuzzy set adopts the Gaussian function: in and The mean and standard deviation are determined by clustering analysis of a large amount of historical normal data. The fuzzy rule set is optimized using an evolutionary algorithm. The optimization objective function is: where y i is the expected output, is the predicted output of the network, K is the number of fuzzy rules L(R k ) is the fuzzy rule R k The complexity penalty term, β is the complexity penalty coefficient, and its value is adjusted through experiments on historical data of different refrigerator brands and models. By continuously evolving the fuzzy rule set, it can accurately learn the vibration mode and characteristics of the door body during normal operation. After receiving the real-time data sample, the data is pattern recognized and analyzed, and the difference between the real-time data and the normal model is calculated to determine whether the current operating state of the door body is normal. Once an abnormality is found, the abnormal data is deeply mined to predict the possible fault type of the door body and the specific time point when the fault may occur; The fault warning and maintenance prompt subsystem: based on the prediction results of the machine learning analysis module, when it is determined that the door body may fail within a specific time period in the future, detailed maintenance prompt information is sent to the user through the visual interface on the refrigerator operation panel, the mobile phone application bound to the refrigerator, and voice prompts; The data storage and management subsystem adopts a hybrid storage architecture, including local non-volatile memory and cloud storage services, to store historical vibration data collected by the vibration data acquisition subsystem, preprocessed data samples, model parameters trained by the machine learning analysis module, and fault prediction information generated by the fault warning and maintenance prompt subsystem, and has data cleaning, optimization and encrypted storage functions, as well as local and cloud data synchronization functions.

2. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The piezoelectric material of the piezoelectric vibration sensor in the vibration data acquisition subsystem is a new type of lead lanthanum zirconate titanate ceramic material, which has a higher piezoelectric constant and a lower mechanical quality factor, and can convert weak mechanical vibrations into stronger electrical signals. The packaging structure of the sensor is specially designed and adopts double-layer protection, with a metal shielding cover on the outer layer and a flexible shock-absorbing material on the inner layer. At the same time, the sensor and the door body are connected by an elastic adhesive, and its elastic modulus is matched according to the elastic modulus of the door body material to ensure good coupling between the sensor and the door body, and can truly reflect the vibration state of the door body. In addition, the installation angle of the sensor is fine-tuned according to the main direction of the door body vibration and the modal analysis results to maximize the reception of vibration energy.

3. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The mixed domain noise suppression algorithm in the data preprocessing module of the data processing and analysis subsystem determines the noise filter h i (t), it is dynamically adjusted according to the different operating modes of the refrigerator. In the refrigeration mode, the operation of the compressor will generate interference of specific frequencies. Through real-time monitoring of the compressor operating frequency, it is taken as the key noise frequency band and a targeted notch filter is designed as h i (t), whose center frequency is the same as the compressor frequency, and the bandwidth is determined according to the half-power bandwidth of the compressor vibration spectrum. For environmental noise, according to the data of the environmental sensor, when the ambient temperature and noise level change, the corresponding filter parameters are automatically adjusted to adapt to the noise characteristics in different environments. At the same time, when calculating the weighting coefficient α i When adjusting the noise, the influence of the number of times the door is opened and closed on the noise is taken into consideration. Every time the door is opened and closed, the noise spectrum is updated in real time according to the speed and force of opening and closing, so as to adjust the weighting coefficient more accurately.

4. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The feature scaling normalization method in the data preprocessing module of the data processing and analysis subsystem assigns different weights to data of different dimensions when calculating normalized data. By analyzing the fault sensitivity of historical data, the weight of amplitude A is determined to be w A =0.4, the weight of frequency F is w F =0.3, phase The weight is The weight of the waveform w(t) is w w =0.

1. This weight distribution is based on the variance contribution rate and correlation analysis of each dimension data under different failure modes. The amplitude shows a higher variance in the door body seal failure failure, so it is given a higher weight. In addition, for the normalized data, data smoothing is further adopted, using a moving average filter: Where M is the smoothing window size, which is determined according to the sampling frequency of the data and the main frequency components of the door vibration to eliminate high-frequency jitter in the data.

5. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The number of input layer nodes of the evolutionary fuzzy neural network in the machine learning analysis module of the data processing and analysis subsystem is determined according to the data dimension and the number of data points in the data segment. For each data segment, it is divided into multiple sub-intervals, each sub-interval corresponds to an input node, and the number of input nodes n=D×S, where D is the data dimension and S is the number of sub-intervals. The sub-intervals are divided based on the typical cycle of the door body vibration and the statistical distribution characteristics of the historical data. In the process of the genetic algorithm generating the fuzzy rule set, the size of the initial population is determined according to the complexity of the refrigerator door body and the diversity of the historical fault data. For the complex door body structure and the diverse fault data, the initial population size is set to a larger value to ensure the diversity of the rule set. In the evolution process, the adaptive crossover probability P is used. c and mutation probability P m , P c and P m Adjust according to the fitness variance of the current population. When the fitness variance is small, increase P c and P m , in order to escape from the local optimal solution.

6. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The visualization interface of the fault warning and maintenance prompt subsystem adopts an ultra-high-definition display screen on the refrigerator operation panel and has a touch interaction function. The interface layout is divided into multiple modules, including a real-time vibration data monitoring module, a fault prediction details module, a maintenance operation guidance module and a user feedback module. The real-time vibration data monitoring module displays various indicators of the current vibration data in the form of dynamic graphics and real-time data lists, and can display the data change trend in the past hour. The fault prediction details module displays the fault risk distribution at different time points in the future in the form of a timeline and a heat map. The darker the color, the higher the fault risk. The maintenance operation guidance module provides detailed illustrated maintenance steps according to the predicted fault type, including the required tools, operation sequence and precautions. For parts that need to be replaced, a three-dimensional model of the parts and an assembly animation demonstration are provided. The user feedback module allows users to evaluate and provide feedback on the maintenance prompts. The system will collect this feedback information to further optimize the fault prediction model and maintenance prompt information.

7. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: When sending information to the mobile phone application, the fault warning and maintenance prompt subsystem will also send augmented reality guidance information. After the user opens the APP, he can aim the mobile phone camera at the refrigerator door and see the virtual fault mark and maintenance operation instructions superimposed on the door. These instructions are accurately displayed according to the predicted fault type and location. The APP also has a remote diagnosis function. When the user turns on this code, the current vibration data and system status information will be sent to the remote server, and professional technicians will provide users with more accurate maintenance suggestions through remote analysis. In addition, the APP supports multi-language services and can automatically switch the display language according to the system language settings of the user's mobile phone. At the same time, it provides user-defined notification settings, allowing users to set reminder times and reminder methods for different fault types.

8. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: The local non-volatile memory of the data storage and management subsystem adopts a three-level storage architecture, which is divided according to the importance and access frequency of the data. The first level stores the latest high-priority data, the second level stores the moderately important data of the past month, and adopts redundant storage technology. The third level stores historical data and adopts compression storage technology to reduce storage space. For cloud storage services, a distributed storage cluster is used. Each storage node allocates storage locations according to the hash value and timestamp of the data, and a consistent hashing algorithm is used to achieve balanced distribution of data. Data encryption adopts multi-key encryption technology, and different encryption keys are used for different types of data. The generation and management of keys are performed through a hardware security module.

9. A refrigerator door vibration sensing detection control system according to claim 1, characterized in that: When the data processing and analysis subsystem conducts in-depth mining of abnormal data to predict the fault type and time point, it also adopts an auxiliary judgment mechanism based on multi-sensor fusion, combines the data of the temperature sensor, humidity sensor and pressure sensor installed inside the refrigerator, and establishes a multi-sensor data fusion model. After normalization and feature extraction of the data of each sensor, the Bayesian reasoning method is used for fusion. The fusion formula is: Where F represents the fault type, D represents the data after multi-sensor fusion, P(F) is the prior probability of the fault type, and P(D|F) is the probability of data D appearing when fault F occurs.

Images