A fault information processing method and system of an intelligent capsule fresh extract beverage machine

Abstract

The application discloses a kind of intelligent capsule fresh extraction beverage machine fault information processing method and system, through the multiple sensors of deployment in each key component of intelligent capsule fresh extraction beverage machine acquisition real-time data, pre-process original sensor data, apply the polynomial time approximation algorithm of complete interval times to carry out multidimensional fault feature analysis and matching, according to fault diagnosis result according to fault severity and user processability grading, generate structured fault information, output fault information through equipment local display interface and mobile terminal, and in cloud server, apply top-down layered differential private counting query mechanism to handle fault diagnosis result, update fault knowledge base.The application realizes the accurate diagnosis, grading processing and multi-channel output of intelligent capsule fresh extraction beverage machine fault, improves fault processing efficiency, and improves user experience.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent household appliance control and data processing, and particularly relates to a fault information processing method and system of an intelligent capsule fresh extract beverage machine. BACKGROUND

[0002] The intelligent capsule fresh extract beverage machine is a product combining modern household appliance technology with artificial intelligence and Internet of Things technology, and belongs to the technical field of intelligent household appliance control and data processing. This kind of device uses special capsule boxes to store raw materials, and combines precisely controlled water temperature, pressure and extraction time to provide users with a convenient high-quality beverage making experience.

[0003] The intelligent beverage machines commonly seen on the market at present mainly use simple single-chip microcomputer control systems, and cooperate with basic sensors (such as temperature, water level, etc.) to realize core functions. For example, a coffee capsule machine of a certain brand detects water temperature through a thermistor, uses a fixed program to control the work of a heating element and a water pump, and only uses a flashing LED light or a simple error code to prompt when a fault occurs.

[0004] Although more advanced beverage machines increase various sensors and microprocessors to monitor more device operation parameters, their fault processing mechanisms still have limitations. This kind of device usually uses a preset fixed fault code corresponding table to make fault judgments, lacks comprehensive analysis ability of multi-sensor data, and leads to insufficiently accurate fault positioning. At the same time, the presentation mode of its fault information is usually limited to simple code display or limited text description, and fails to fully utilize data processing technology to provide more intelligent fault analysis and solution suggestions.

[0005] This traditional fault processing mechanism has obvious deficiencies: first, the fault information indication is not intuitive and explicit enough, and ordinary users cannot understand the specific fault reasons; second, it lacks targeted fault solution guidance, which leads to the fact that users cannot self-troubleshoot simple faults; third, fault data cannot be effectively collected and analyzed, which leads to low after-sales maintenance efficiency; fourth, it cannot realize remote diagnosis and preventive maintenance, which increases the cost of device maintenance. SUMMARY

[0006] The present application aims to overcome the deficiencies in the prior art, and provides a fault information processing method and system of an intelligent capsule fresh extract beverage machine, which realizes accurate diagnosis, hierarchical processing and multi-channel output of faults, improves fault processing efficiency, and improves user experience.

[0007] To achieve the above-mentioned purpose, the present application provides a fault information processing method of an intelligent capsule fresh extract beverage machine, which comprises:

[0008] The raw sensor data is obtained by collecting real-time data of temperature, pressure, flow, current, voltage and vibration frequency through a plurality of sensors deployed on each key component of the intelligent capsule fresh extract beverage machine.

[0009] The raw sensor data is preprocessed by filtering and normalization to obtain preprocessed sensor data.

[0010] Based on the preprocessed sensor data, a complete interval polynomial time approximation algorithm is applied for multi-dimensional fault feature analysis and matching to obtain a fault diagnosis result.

[0011] According to the fault diagnosis result, the fault is classified according to the severity and user's processability, and structured fault information including fault description, cause analysis and solution is generated.

[0012] The structured fault information is output through the local display interface of the device and the mobile terminal, and the fault diagnosis result is uploaded to the cloud server.

[0013] In the cloud server, a top-down hierarchical differential private counting query mechanism is applied to process the fault diagnosis result, and the fault knowledge base is updated.

[0014] Preferably, the raw sensor data is obtained by collecting real-time data of temperature, pressure, flow, current, voltage and vibration frequency through a plurality of sensors deployed on each key component of the intelligent capsule fresh extract beverage machine, including:

[0015] Based on the preset sampling time interval and sampling trigger condition, the analog signals of each sensor are sampled at regular intervals to obtain time series sampling data.

[0016] The time series sampling data is converted into digital sensor data by analog-to-digital conversion.

[0017] The digital sensor data is transmitted to the main control unit through the internal bus, and a time stamp is added to the received sensor data in the main control unit to obtain time-stamped sensor data.

[0018] In the main control unit, a data buffer is established, and the time-stamped sensor data is indexed according to sensor type and collection time to obtain a structured raw sensor data set.

[0019] Preferably, the raw sensor data is preprocessed by filtering and normalization to obtain preprocessed sensor data, including:

[0020] The structured raw sensor data set is applied to median filtering and Kalman filtering algorithm to eliminate noise interference and obtain filtered data.

[0021] normalizing the filtered data, mapping different types of data to a unified interval, to obtain standardized data;

[0022] Based on historical operation data statistical analysis, the normal value interval and fluctuation threshold of each sensor data are established, and the parameter reference data set is obtained;

[0023] The standardized data is compared with the parameter reference data set, and the abnormal data points are marked according to the preset deviation threshold, to obtain the pretreated sensor data with abnormal marks.

[0024] Preferably, based on the pretreated sensor data, a complete interval frequency polynomial time approximation algorithm is applied for multi-dimensional fault feature analysis and matching, to obtain a fault diagnosis result, including:

[0025] Collect and organize historical fault cases, extract fault feature indicators and corresponding sensor data features, and obtain a fault feature mapping table;

[0026] Based on the fault feature mapping table, a fault feature library containing fault types, sensor data features and occurrence scenarios is established, to obtain fault feature reference data;

[0027] The pretreated sensor data with abnormal marks is divided into intervals, the data distribution density of each interval is calculated, and an interval frequency vector is obtained;

[0028] The interval frequency vector is compared with the fault feature reference data to obtain a fault type candidate set;

[0029] The fault type candidate set is analyzed for component association to determine the fault occurrence position and impact range, to obtain a fault diagnosis result.

[0030] Preferably, the pretreated sensor data with abnormal marks is divided into intervals, the data distribution density of each interval is calculated, and an interval frequency vector is obtained, including:

[0031] The value range of each sensor data is equally divided to establish a numerical interval mapping table, and an interval division scheme is obtained;

[0032] Based on the interval division scheme, the data distribution frequency of each interval is counted, and the interval frequency vector is generated;

[0033] The interval frequency vector is compared with the fault feature reference data to obtain a fault type candidate set, including:

[0034] A multi-layer neural network model is constructed to extract features from the interval frequency vector, to obtain a fault feature representation;

[0035] a cosine similarity between the fault feature representation and a preset pattern in the fault feature reference data is calculated to obtain a similarity matrix;

[0036] based on the similarity matrix, a fault type with a similarity exceeding a preset threshold is screened to obtain a preliminary fault type;

[0037] a decision tree is constructed based on the preliminary fault type and the fault preliminary judgment results of the plurality of sensors to determine the fault type to obtain a fault type candidate set.

[0038] Preferably, according to the fault diagnosis result, the fault is graded according to the fault severity and user handleability to generate structured fault information containing fault description, cause analysis and solution, including:

[0039] a fault evaluation index system is established according to the degree of influence of the device function, the safety risk level and the maintenance difficulty to obtain an evaluation benchmark;

[0040] the fault diagnosis result is quantitatively scored based on the evaluation benchmark to divide the fault level to obtain a fault grading result;

[0041] a description template corresponding to the fault grading result is extracted from a preset fault description template library, and fault parameters and influence range are filled to obtain fault description content;

[0042] a probability distribution of different fault causes is calculated based on statistical analysis of historical fault data to generate cause analysis content;

[0043] a corresponding handling plan is selected according to the fault grading result, and inspection, operation and verification steps are generated according to the device state to obtain solution content;

[0044] the fault description content, the cause analysis content and the solution content are integrated into a unified format to obtain structured fault information.

[0045] Preferably, the structured fault information is output through a device local display interface and a mobile terminal, and the fault diagnosis result is uploaded to a cloud server, including:

[0046] the structured fault information is divided into emergency alarm information, fault prompt information and maintenance suggestion information according to the fault level to obtain a hierarchical information set;

[0047] based on the hierarchical information set, a basic information layout of the local display interface and a detailed information layout of the mobile terminal are configured to obtain a display configuration scheme;

[0048] according to the display configuration scheme, a fault name, an alarm level and a basic handling prompt are output on the device local display interface, and the alarm level is output through different colors and flashing frequencies of LED indicator lights.

[0049] The structured fault information is transmitted to the user's mobile terminal via the MQTT protocol. The fault diagnosis process, cause analysis, and handling solutions are displayed in layers according to the importance of the information, and the mobile terminal output information is obtained.

[0050] The fault diagnosis results are encapsulated in JSON format and encrypted with AES, and then uploaded to the cloud server via WiFi or Bluetooth communication.

[0051] Preferably, a top-down hierarchical differential private counting query mechanism is applied in the cloud server to process the fault diagnosis results and update the fault knowledge base, including:

[0052] The cloud server receives and decrypts the fault diagnosis results uploaded by multiple devices, establishes a structured fault data table, and obtains the original fault statistics data.

[0053] The original fault statistics data are subjected to hierarchical differential processing to calculate the frequency of fault occurrence while protecting user privacy, and to obtain fault type distribution statistics.

[0054] Based on the statistical distribution of the aforementioned fault types, combined with equipment runtime and usage environment parameters, the timing characteristics and environmental correlation of faults are analyzed to obtain a fault mode analysis report.

[0055] Based on the fault mode analysis report, update the feature weights and threshold parameters of the fault feature library, optimize the fault diagnosis rules, and obtain an improved diagnostic model.

[0056] By integrating user feedback and maintenance records, extracting new fault characteristics and solutions, and updating the fault knowledge base, an iteratively optimized fault knowledge base is obtained.

[0057] Preferably, the original fault statistics are subjected to hierarchical differential processing to calculate the frequency of fault occurrence while protecting user privacy, thereby obtaining fault type distribution statistics, including:

[0058] The original fault statistics data are indexed in multiple layers according to equipment model, usage time, geographical region and ambient temperature to obtain a hierarchical data structure.

[0059] In each layer of data index, fault records are grouped and counted, and Laplace noise is added according to the differential privacy budget to obtain the privatized counting result;

[0060] Based on the privatized counting results, the time distribution characteristics of the fault occurrence frequency are calculated using a sliding time window to obtain fault trend data.

[0061] Hierarchical clustering analysis is performed on the fault trend data to generate a fault frequency heatmap and a time-series trend map, resulting in visual statistical charts;

[0062] By combining equipment operating parameters and environmental monitoring data, correlation analysis is used to identify the causes of failures and obtain a statistical report on the distribution of failure types.

[0063] The present invention also provides a fault information processing device for an intelligent capsule fresh-extract beverage machine, comprising:

[0064] The data acquisition module is used to collect real-time data on temperature, pressure, flow rate, current, voltage and vibration frequency through multiple sensors deployed in key components of the intelligent capsule fresh-extract beverage machine, and obtain raw sensor data.

[0065] The data preprocessing module is used to filter and normalize the raw sensor data to obtain preprocessed sensor data.

[0066] The fault diagnosis module is used to perform multidimensional fault feature analysis and matching based on the preprocessed sensor data, using a polynomial-time approximation algorithm of complete interval degree, to obtain fault diagnosis results.

[0067] The information generation module is used to classify the faults according to their severity and user manageability based on the fault diagnosis results, and generate structured fault information that includes fault description, cause analysis and solutions.

[0068] The information output module is used to output the structured fault information through the local display interface of the device and the mobile terminal, and to upload the fault diagnosis results to the cloud server.

[0069] The data processing module is used to process the fault diagnosis results and update the fault knowledge base by applying a top-down hierarchical differential private counting query mechanism in the cloud server.

[0070] The beneficial effects of this invention are:

[0071] A multinomial-time approximation algorithm of complete interval degree is applied to perform multidimensional fault feature analysis, which enables rapid and accurate identification of complex fault modes and significantly improves the accuracy and efficiency of fault diagnosis.

[0072] A top-down hierarchical differential private counting query mechanism is introduced for fault data processing. While protecting user data privacy, it achieves efficient fault data statistical analysis and provides data support for continuous optimization of fault handling.

[0073] A fault diagnosis system based on multi-sensor data fusion was constructed, and the accurate location and classification of faults were achieved through comprehensive analysis of multi-dimensional parameters.

[0074] A fault classification and handling mechanism has been established, which generates differentiated fault information and solutions based on the severity of the fault and the user's ability to handle it, thereby improving the pertinence and efficiency of fault handling.

[0075] A multi-channel fault information output method was designed to meet the user's needs for obtaining fault information in different scenarios and improve the user experience;

[0076] A self-learning fault knowledge base update mechanism was created, which enables continuous optimization and iterative upgrades of fault handling through the analysis of user feedback and maintenance records. Attached Figure Description

[0077] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0078] Figure 1 A flowchart illustrating a fault information processing method for an intelligent capsule fresh-extract beverage machine provided in an embodiment of the present invention;

[0079] Figure 2 This is a flowchart of real-time data acquisition using multiple sensors provided in an embodiment of the present invention;

[0080] Figure 3 This is a flowchart of data preprocessing and anomaly detection provided in an embodiment of the present invention;

[0081] Figure 4 This is a flowchart of multi-dimensional fault diagnosis and location provided in an embodiment of the present invention;

[0082] Figure 5 A flowchart for generating graded fault information provided in an embodiment of the present invention;

[0083] Figure 6 This is a structural diagram of a fault information processing system for an intelligent capsule fresh-extract beverage machine provided in an embodiment of the present invention. Detailed Implementation

[0084] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0085] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0086] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0087] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0088] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0089] like Figure 1 As shown, the present invention provides a fault information processing method for an intelligent capsule fresh-extract beverage machine, comprising the following steps:

[0090] Step S1: Collect real-time data on temperature, pressure, flow rate, current, voltage, and vibration frequency by using multiple sensors deployed on key components of the smart capsule fresh-extract beverage machine to obtain raw sensor data.

[0091] In this embodiment, as Figure 2 As shown, step S1 specifically includes:

[0092] Step S1.1: Based on the preset sampling time interval and sampling trigger conditions, the analog signals of each sensor are sampled at regular intervals to obtain time-series sampling data.

[0093] Different types of sensors are installed on key components of the intelligent capsule fresh-extract beverage machine, including temperature sensors, pressure sensors, flow sensors, current sensors, voltage sensors, and vibration sensors. Different sampling time intervals are set for different types of sensors; for example, the temperature sensor samples every 5 seconds, the pressure sensor every 2 seconds, and the flow sensor every 1 second. Simultaneously, sampling trigger conditions are set, such as triggering sampling when the device starts up, its operating status changes, or when the user operates the device. In this way, time-series sampling data from each sensor is acquired.

[0094] Step S1.2: Perform analog-to-digital conversion on the time-series sampling data to obtain digital sensor data.

[0095] The acquired analog signals are converted into digital signals using an ADC (Analog-to-Digital Converter). For example, for a temperature sensor, an analog voltage signal of 0-5V is converted into a digital value of 0-1023; for a pressure sensor, a current signal of 4-20mA is converted into a corresponding digital value. In this way, the sensor data is digitally processed.

[0096] Step S1.3: Transmit the digital sensor data to the main control unit via the internal bus, and add a timestamp to the received sensor data in the main control unit to obtain the timestamped sensor data.

[0097] Digital sensor data is transmitted to the main control unit via internal bus protocols such as I²C and SPI. In the main control unit, a timestamp is added to each set of sensor data to record the precise time of data acquisition, facilitating subsequent time-series analysis and data correlation.

[0098] Step S1.4: Establish a data cache in the main control unit, and index the timestamp-marked sensor data according to sensor type and acquisition time to obtain a structured raw sensor data group.

[0099] A data cache area is established in the memory of the main control unit to organize and manage the received sensor data. An index structure is established according to sensor type (such as temperature, pressure, flow rate, etc.) and acquisition time to form a structured raw sensor data set, which facilitates subsequent data processing and analysis.

[0100] Step S2: Perform filtering and normalization preprocessing on the original sensor data to obtain preprocessed sensor data.

[0101] In this embodiment, as Figure 3 As shown, step S2 specifically includes:

[0102] Step S2.1: Apply mean filtering and Kalman filtering algorithms to the structured raw sensor data set to eliminate noise interference and obtain filtered data.

[0103] Apply a mean-mode filtering algorithm to the raw sensor data to remove sudden noise and outliers. For example, for temperature sensor data, the median of five consecutive sampling points is used as the filtering result. For data requiring smoothing, such as pressure and flow data, a Kalman filter algorithm is applied, dynamically adjusting the filtering parameters based on historical data and the uncertainty of current measurements to obtain a smoother data curve.

[0104] Step S2.2: Normalize the filtered data to map different types of data to a unified interval to obtain standardized data.

[0105] Different types of sensor data are normalized and mapped to the [0,1] interval. For example, temperature data can be normalized using the formula (T-Tmin) / (Tmax-Tmin), where T is the current temperature value, and Tmin and Tmax are the preset minimum and maximum temperature values, respectively. This method unifies the scale of data from different sensors, facilitating subsequent comprehensive analysis.

[0106] Step S2.3: Based on the statistical analysis of historical operating data, establish the normal value range and fluctuation threshold of each sensor data to obtain the parameter benchmark dataset.

[0107] By analyzing historical data from normal equipment operation, the statistical characteristics of each sensor's data are calculated, such as mean, standard deviation, maximum, and minimum values. Based on these statistical characteristics, normal value ranges and fluctuation thresholds for each sensor's data are established. For example, the normal value range for a temperature sensor might be [85℃, 95℃], with a fluctuation threshold of ±2℃. These parameters constitute a parameter benchmark dataset, serving as the basis for anomaly detection.

[0108] Step S2.4: Compare the standardized data with the parameter benchmark dataset, mark abnormal data points according to the preset deviation threshold, and obtain preprocessed sensor data with abnormal markings.

[0109] The standardized sensor data is compared with a parameter benchmark dataset, and the deviation value is calculated. If the deviation value exceeds a preset threshold, it is marked as an abnormal data point. For example, if the standardized value of the temperature sensor deviates from the benchmark value by more than 0.1 (corresponding to an actual temperature of approximately ±5℃), it is marked as a temperature anomaly. In this way, preprocessed sensor data with anomaly markings is obtained, providing a basis for subsequent fault diagnosis.

[0110] Step S3: Based on the preprocessed sensor data, apply a multidimensional fault feature analysis and matching algorithm of full interval degree to obtain fault diagnosis results.

[0111] In this embodiment, as Figure 4 As shown, step S3 specifically includes:

[0112] Step S3.1: Collect and organize historical fault cases, extract fault characteristic indicators and corresponding sensor data characteristics, and obtain a fault characteristic mapping table.

[0113] The system systematically collects various potential fault information for intelligent capsule fresh-extract beverage machines. This includes methods such as laboratory simulation testing, analysis of historical maintenance records, and summaries of expert experience to comprehensively identify possible fault types, such as abnormal heating, water pump failure, communication interruption, and capsule recognition errors. For each fault type, unique sensor data characteristic patterns are extracted. For example, when the heating module malfunctions, the temperature sensor may display abnormal temperature curves (such as excessively slow heating or excessive temperature fluctuations); when the water pump is clogged, the pressure and flow sensor data may both show abnormalities. In this way, a fault feature mapping table is established to provide a reference for subsequent fault diagnosis.

[0114] Step S3.2: Based on the fault feature mapping table, establish a fault feature library containing fault type, sensor data features and occurrence scenario to obtain fault feature reference data.

[0115] Based on the fault feature mapping table, a structured fault feature knowledge base is established, creating a clear mapping relationship between each fault type and its corresponding sensor data feature patterns. This knowledge base includes information such as fault type ID, fault name, involved sensor type, sensor data feature description (e.g., numerical range, trend of change, frequency characteristics, etc.), and typical scenarios in which the fault occurs. This knowledge base serves as the foundation for fault identification and provides a reference for subsequent fault diagnosis algorithms.

[0116] Step S3.3: Divide the preprocessed sensor data with anomaly markers into intervals, calculate the data distribution density of each interval, and obtain the interval frequency vector.

[0117] The preprocessed sensor data with anomaly markers is divided into intervals, and the data distribution density of each interval is calculated to obtain the interval exponent vector, specifically including:

[0118] Step S3.3.1: Divide the range of values ​​of each sensor data into equal intervals, establish a numerical interval mapping table, and obtain the interval division scheme.

[0119] The data range of each sensor is divided into intervals. For example, the 0-100℃ range of the temperature sensor is divided into 10 intervals, each with a width of 10℃; the 0-10bar range of the pressure sensor is divided into 5 intervals, each with a width of 2bar. In this way, a numerical interval mapping table is established as the basis for interval frequency statistics.

[0120] Step S3.3.2: Based on the interval division scheme, count the data distribution frequency of each interval and generate the interval frequency vector.

[0121] The frequency of each sensor's data falling into each interval within a certain period (such as the last 10 minutes) is statistically analyzed to form an interval frequency vector. For example, the interval frequency vector of a temperature sensor might be [0,0,0,0,0,0,0,15,85,0], indicating that in the last 10 minutes, the temperature was in the 70-80℃ range for 15% of the time and in the 80-90℃ range for 85% of the time.

[0122] Step S3.4: Calculate the similarity between the interval frequency vector and the fault feature reference data to obtain a fault type candidate set.

[0123] The similarity between the interval frequency vector and the fault feature reference data is calculated to obtain a candidate set of fault types, which specifically includes:

[0124] Step S3.4.1: Construct a multi-layer neural network model, extract features from the interval frequency vector, and obtain the fault feature representation.

[0125] Construct a multi-layer neural network model to extract features from interval frequency vectors. This model can be a simple feedforward neural network containing an input layer, hidden layers, and an output layer. The input layer receives the interval frequency vector, and through nonlinear transformations in the hidden layers, the output layer generates a fault feature representation. This feature representation can capture complex patterns in the interval frequency vector, facilitating subsequent similarity calculations.

[0126] In the fault diagnosis process of intelligent capsule fresh-extract beverage machines, the construction and training of a multi-layer neural network model is a key step in achieving high-precision fault feature extraction. This model is mainly used to transform interval frequency vectors into more expressive fault feature representations. The construction and training process is described in detail below.

[0127] First, the basic architecture of the model adopts a three-layer feedforward neural network structure, including an input layer, hidden layers, and an output layer. The number of neurons in the input layer matches the dimension of the interval frequency vector. For example, for six types of sensor data—temperature, pressure, flow rate, current, voltage, and vibration frequency—if each type of sensor data is divided into 10 intervals, the input layer has 60 neurons. The hidden layer uses a two-layer structure: the first hidden layer contains 128 neurons, and the second hidden layer contains 64 neurons. The output layer has a dimension of 32, generating a 32-dimensional fault feature representation vector.

[0128] The choice of activation function is crucial to model performance. Using the ReLU (Rectified LinearUnit) activation function in the hidden layers, expressed as f(x) = max(0,x), helps solve the vanishing gradient problem in deep network training and accelerates the convergence process. The output layer uses the tanh activation function, restricting the output values ​​to the range [-1,1], which facilitates subsequent similarity calculations.

[0129] The model is trained using supervised learning, employing labeled historical fault data as the training set. The training data includes a vector of the number of fault occurrences within a given interval as input, and the corresponding fault type as the label. To enhance the model's generalization ability, the training set contains fault cases under various operating conditions, such as fault data under different ambient temperatures and usage frequencies. A typical training set size is approximately 5000-10000 samples, covering 30-50 common fault types.

[0130] The optimizer chosen is Adam (Adaptive Moment Estimation), which combines the advantages of momentum and RMSProp, and can adaptively adjust the learning rate. The initial learning rate is set to 0.001. The loss function used is cross-entropy loss, which is suitable for multi-class classification problems. During training, a batch size of 64 and 100 epochs are used, and an early stopping strategy is employed: training stops when the loss on the validation set fails to improve for 10 consecutive epochs to prevent overfitting.

[0131] To further improve model performance, the following techniques were also employed:

[0132] Data augmentation: Gaussian noise (μ=0, σ=0.01) is added to the original interval exponent vector to simulate small fluctuations in sensor data and enhance model robustness.

[0133] Dropout regularization: Add a dropout layer between two hidden layers with a dropout rate of 0.3 to reduce the risk of overfitting.

[0134] Batch normalization: Add a batch normalization layer after each hidden layer to accelerate the training process and improve model stability.

[0135] After training, the model's performance is evaluated using a test set, with typical accuracy reaching 92%-95%. In practical applications, this model can transform the original interval frequency vector into a more discriminative feature representation. For example, for a "water pump blockage" fault, its feature representation will show obvious activation patterns in specific dimensions, which is significantly different from the feature representation of a "heater fault," thus providing a reliable foundation for subsequent similarity calculations and fault diagnosis.

[0136] Step S3.4.2: Calculate the cosine similarity between the fault feature representation and the preset pattern in the fault feature reference data to obtain a similarity matrix.

[0137] The cosine similarity between the fault feature representation generated by the neural network and the preset patterns in the fault feature database is calculated. The formula for calculating the cosine similarity is cos(θ) = A·B / (|A|·|B|), where A and B are the current feature vector and the reference feature vector, respectively. In this way, the similarity matrix between the current fault features and various preset fault patterns is obtained.

[0138] Step S3.4.3: Based on the similarity matrix, filter out fault types with similarity exceeding a preset threshold to obtain preliminary fault types.

[0139] Set a similarity threshold (e.g., 0.8) and filter out fault types with similarity exceeding this threshold as preliminary fault types. For example, if the current feature has a similarity of 0.92 with the "water pump blockage" pattern and a similarity of 0.75 with the "heater failure" pattern, it is initially judged as a "water pump blockage" fault.

[0140] Step S3.4.4: Combine the initial fault judgment results of multiple sensors to construct a decision tree to determine the fault type and obtain a candidate set of fault types.

[0141] Based on the preceding feature analysis results, a multi-level decision tree structure is designed to achieve accurate classification of fault types through a series of conditional judgments. Each node of the decision tree represents a judgment condition (such as "whether the temperature exceeds 85℃", "whether the flow rate is lower than 50ml / min", etc.), each branch represents the result of the conditional judgment, and each leaf node corresponds to a fault type or the next level of decision.

[0142] When designing a decision tree, the optimal decision order is determined based on prior knowledge of the fault and the correlation of sensor data. Typically, the decision begins with the most discriminative feature and gradually refines. For example, first determine which system module is malfunctioning (e.g., heating system, water pump system), and then further determine the specific fault type. This multi-level decision logic can handle complex fault situations, including cases where multiple abnormal parameters occur simultaneously.

[0143] Step S3.5: Perform component association analysis on the candidate set of fault types to determine the location and scope of the fault and obtain the fault diagnosis results.

[0144] After determining the type of fault, the next step is to pinpoint the specific location and extent of the fault. First, a topological model of the relationships between the various components of the equipment is established, describing the physical connections and functional dependencies between them. For example, there is a water flow transmission relationship between the water pump and the heater, and a temperature sensing relationship between the heater and the temperature sensor.

[0145] Based on this topology model, when an anomaly is detected in a component, the potential impact on its upstream and downstream components, as well as possible propagation paths of the fault, are analyzed. For example, when insufficient water flow is detected, possible fault points include pump failure, pipe blockage, or water tank depletion. By analyzing the time-series changes and interrelationships of data from various monitoring points, the fault range can be further narrowed down. This component correlation analysis can not only accurately locate the fault position but also assess the scope of its impact, providing important reference for subsequent fault handling.

[0146] Step S4: Based on the fault diagnosis results, classify the faults according to their severity and user manageability, and generate structured fault information that includes fault description, cause analysis and solutions.

[0147] In this embodiment, as Figure 5 As shown, step S4 specifically includes:

[0148] Step S4.1: Based on the degree of impact on equipment function, safety risk level, and maintenance difficulty, establish a fault assessment index system to obtain the assessment benchmark.

[0149] Establish a systematic fault classification standard to assess and classify faults from multiple dimensions. First, consider the degree of impact of the fault on the equipment function, and divide it into: minor faults that do not affect use (such as failure of some non-core indicator lights), general faults that partially affect the function (such as unstable extraction temperature), serious faults that render the main functions unusable (such as inability to heat), and dangerous faults that may cause safety hazards (such as excessive pressure, short circuit, etc.).

[0150] Secondly, the hazard level of the malfunction is assessed from a safety risk perspective, including no safety risk, low risk (may affect beverage quality), medium risk (may cause equipment damage), and high risk (may cause personal injury or fire, etc.). Thirdly, from a user-manageability perspective, malfunctions are categorized into simple malfunctions that users can resolve themselves (such as lack of water, capsule not placed, etc.), medium malfunctions requiring basic maintenance (such as simple cleaning or resetting), and complex malfunctions requiring professional repair (such as damage to internal components).

[0151] These hierarchical dimensions combine to form a multi-dimensional fault classification matrix, assigning a comprehensive level to each fault to guide subsequent information generation and processing. For example, an "A1 level fault" might represent a fault type that "affects major functions and poses security risks, but can be handled by the user."

[0152] Step S4.2: Quantitatively score the fault diagnosis results based on the evaluation criteria, classify the fault levels, and obtain the fault classification results.

[0153] The fault diagnosis results are quantitatively scored, including factors such as functional impact (1-5 points), safety risk level (1-5 points), and repair difficulty (1-5 points). Based on the scoring results, the fault is classified into different levels, for example:

[0154] 1-3 points: Minor fault, which the user can handle themselves;

[0155] 4-6 points: General fault, requires basic maintenance;

[0156] 7-9 points: Serious malfunction, requires professional repair;

[0157] 10-15 points: Dangerous malfunction, require immediate shutdown and contact of professionals.

[0158] In this way, fault classification results are obtained, providing a basis for subsequent information generation.

[0159] Step S4.3: Extract the description template corresponding to the fault classification result from the preset fault description template library, fill in the fault parameters and the scope of influence, and obtain the fault description content.

[0160] Based on the identified fault types, corresponding templates are extracted from a pre-set fault description template library. These templates contain three main parts: fault name (such as "water pump blockage", "heater overheat protection", etc.), fault symptoms (such as "no water flow during production", "abnormally prolonged preheating time", etc.), and related technical parameters (such as "water flow rate is less than 10ml / min", "temperature exceeds 95℃ for 30 seconds", etc.).

[0161] Specific parameter values ​​are filled into the template, such as "Current water temperature is 87℃, lower than the normal operating temperature of 92℃". Descriptions of varying complexity can be generated for different user groups (such as general users and professional repair personnel) to meet their diverse information needs.

[0162] Step S4.4: Based on the statistical analysis of historical fault data, calculate the probability distribution of different fault causes and generate cause analysis content.

[0163] The system extracts relevant possible causes from a pre-established fault cause knowledge base. This knowledge base is built based on historical fault data, expert experience, and theoretical analysis, and includes common causes of various faults and their probabilities of occurrence. For example, for the fault "the water pump cannot work properly," possible causes include: water pump motor damage (30%), control circuit failure (25%), water pipe blockage (35%), power supply problems (10%), etc.

[0164] Based on the characteristics of the monitored data, various possible causes are ranked by probability, and the most likely causes are presented to the user first. Simultaneously, a detailed explanation is provided for each possible cause to help users understand the mechanism of the fault. This probability-based cause analysis can more effectively guide users or maintenance personnel in troubleshooting, improving the efficiency of fault resolution.

[0165] Step S4.5: Select the corresponding handling plan based on the fault classification result, and combine the equipment status generation inspection, operation and verification steps to obtain the solution content.

[0166] Based on the type, level, and possible causes of the malfunction, corresponding solution suggestions are generated for the user. For malfunctions that the user can handle themselves (such as water shortage, incorrect capsule placement, etc.), detailed operating procedure instructions are provided, including textual descriptions and possible illustrative references. For example, for a "water tank low" malfunction, the solution could be: "1. Remove the water tank; 2. Add clean drinking water to the tank up to the MAX mark; 3. Reinstall the water tank onto the device; 4. Press the retry button."

[0167] For malfunctions requiring basic maintenance, simple maintenance instructions are provided, such as cleaning procedures and reset methods. For complex malfunctions requiring professional repair, suggestions to contact customer service or report the problem are offered, and some temporary usage restrictions may be included to prevent further damage to the equipment. For malfunctions that may pose safety risks, safety precautions are emphasized, such as warnings like "Please disconnect the power immediately and do not attempt to repair it yourself."

[0168] Step S4.6: Integrate the fault description, cause analysis, and solution into a unified format to obtain structured fault information.

[0169] The fault description, cause analysis, and solutions generated in the previous steps are combined to form a complete structured fault information content. This structured content contains a clear hierarchical structure, starting with the fault name and brief description, moving to possible cause analysis, and then to specific solution suggestions. Furthermore, the information content is formatted accordingly based on the characteristics of different output channels (such as device displays, mobile apps, etc.).

[0170] For example, a device's local display screen, due to limited display space, may only show a brief fault name and basic troubleshooting steps; while a mobile app can display complete fault information, including detailed cause analysis, illustrated solution guidance, and possibly even links to video tutorials. Furthermore, different visual styles (such as color coding, icons, etc.) can be used to differentiate faults of varying severity, enhancing the user's intuitive understanding of the fault's severity.

[0171] Step S5: Output the structured fault information through the device's local display interface and mobile terminal, and upload the fault diagnosis results to the cloud server.

[0172] In this embodiment, step S5 specifically includes:

[0173] Step S5.1: Divide the structured fault information into emergency alarm information, fault prompt information and maintenance suggestion information according to the fault level to obtain a hierarchical information set.

[0174] Based on the severity and urgency of the fault, structured fault information is divided into different levels:

[0175] Emergency alarm information: High-risk faults that require immediate user intervention, such as excessive pressure or abnormal temperature;

[0176] Fault message: Faults that affect the normal use of the equipment but do not pose a safety risk, such as low water pump efficiency or capsule recognition error;

[0177] Maintenance recommendations: Tips that do not affect current use but require regular maintenance, such as recommendations to clean the water system or replace the filter.

[0178] This hierarchical approach creates a tiered information set, providing a foundation for information output through different channels.

[0179] Step S5.2: Based on the hierarchical information set, configure the basic information layout of the local display interface and the detailed information layout of the mobile terminal to obtain the display configuration scheme.

[0180] Design corresponding display layouts based on different levels of fault information:

[0181] For emergency alarm information, use a striking red background and warning icon on the local display interface, and push notifications on mobile devices and display them at the top.

[0182] For fault prompts, a yellow background and prompt icon are used on the local display interface, while they are displayed in the fault list on the mobile terminal.

[0183] Maintenance recommendations are displayed in the information area of ​​the local display interface and on the maintenance recommendations page of the mobile terminal.

[0184] In this way, display configuration schemes for different devices can be obtained, ensuring the effective transmission of information.

[0185] Step S5.3: According to the display configuration scheme, output the fault name, alarm level and basic processing prompts on the local display interface of the device, and output the alarm level through different colors and flashing frequencies of LED indicator lights.

[0186] On the device's local display interface, fault information is output according to the display configuration scheme. For example, for a "water pump blockage" fault, it may display "Fault: Water pump blockage [Medium]" and provide a brief handling prompt "Please check if the water pipe is blocked". At the same time, different states of LED indicator lights indicate different alarm levels, such as rapid red flashing for an emergency alarm, slow yellow flashing for a general fault, and solid green for a normal state.

[0187] Step S5.4: Transmit the structured fault information to the user's mobile terminal via the MQTT protocol, and display the fault diagnosis process, cause analysis and handling plan in layers according to the importance of the information, so as to obtain the mobile terminal output information.

[0188] Structured fault information is transmitted to the user's mobile app via the MQTT (Message Queuing Telemetry Transport) protocol. Within the app, the information is displayed in layers according to its importance.

[0189] The homepage displays the fault name, severity level, and brief description;

[0190] The details page displays the fault diagnosis process, possible cause analysis, and solutions;

[0191] The advanced information page provides technical specifications, historical records, and links to relevant knowledge.

[0192] This layered display method satisfies the different users' needs for information depth and improves the user experience.

[0193] Step S5.5: Encapsulate the fault diagnosis results in JSON format and encrypt them with AES, then upload them to the cloud server via WiFi or Bluetooth communication.

[0194] The fault diagnosis results are converted into JSON format, containing information such as device ID, fault type, fault time, sensor data, and diagnostic results. The data is then encrypted using the AES encryption algorithm to ensure secure data transmission. Finally, the encrypted data is uploaded to a cloud server via WiFi or Bluetooth for subsequent data analysis and knowledge base updates.

[0195] Step S6: Apply a top-down hierarchical differential private counting query mechanism to process the fault diagnosis results on the cloud server and update the fault knowledge base.

[0196] In this embodiment, step S6 specifically includes:

[0197] Step S6.1: The cloud server receives and decrypts the fault diagnosis results uploaded by multiple devices, establishes a structured fault data table, and obtains the original fault statistics data.

[0198] The cloud server receives encrypted fault diagnosis data from multiple devices and decrypts it using the corresponding decryption keys. The decrypted data is then organized into a structured fault data table, containing fields such as device ID, fault type, fault time, sensor data, and diagnostic results. This process yields raw fault statistics, providing a foundation for subsequent analysis.

[0199] Step S6.2: Perform hierarchical differential processing on the original fault statistics data, calculate the frequency of fault occurrence under the premise of protecting user privacy, and obtain fault type distribution statistics.

[0200] The raw fault statistics are subjected to stratified differential processing. While protecting user privacy, the frequency of fault occurrence is calculated to obtain fault type distribution statistics, specifically including:

[0201] Step S6.2.1: Establish a multi-level index for the original fault statistics data according to equipment model, usage time, geographical region and ambient temperature to obtain a hierarchical data structure.

[0202] The raw fault statistics data are indexed hierarchically according to multiple dimensions, including device model (e.g., basic version, advanced version), usage duration (e.g., 0-6 months, 6-12 months), geographical region (e.g., North China, East China), and ambient temperature (e.g., low temperature, normal temperature, high temperature). This multi-dimensional indexing forms a hierarchical data structure, facilitating subsequent hierarchical analysis.

[0203] Step S6.2.2: In each layer of data index, count the fault records in groups and add Laplace noise according to the differential privacy budget to obtain the privatized count result.

[0204] Within each level of the data index, fault records are grouped and counted according to fault type. To protect user privacy, differential privacy technology is applied, adding random noise conforming to a Laplace distribution to the counting results. The magnitude of the noise is controlled by the differential privacy budget ε; a smaller ε value provides stronger privacy protection but reduces data accuracy. In this way, privatized counting results are obtained, preserving statistical significance while protecting individual user privacy.

[0205] Step S6.2.3: Based on the privatized counting results, the time distribution characteristics of the fault occurrence frequency are calculated using a sliding time window to obtain fault trend data.

[0206] A time-series analysis of the privatization count results is performed using a sliding time window (e.g., 30 days) to calculate the frequency of various types of faults within different time periods. By comparing data from different time windows, trends in fault frequency can be identified, such as whether a certain type of fault shows an upward trend or whether there are seasonal fluctuations. This trend data provides important information for fault prediction and preventative maintenance.

[0207] Step S6.2.4: Perform hierarchical cluster analysis on the fault trend data to generate a fault frequency heatmap and a time series trend map, and obtain visual statistical charts.

[0208] Hierarchical clustering analysis is performed on fault trend data to identify fault types and equipment groups with similar patterns. Based on the clustering results, a fault frequency heatmap is generated to visually display the frequency of various faults occurring under different equipment types and operating environments; simultaneously, a time-series trend chart is generated to show the changing trend of fault frequency over time. These visual statistical charts provide intuitive support for fault analysis and decision-making.

[0209] Step S6.2.5: Combining equipment operating parameters and environmental monitoring data, identify fault causes through correlation analysis and obtain a statistical report on the distribution of fault types.

[0210] By correlating fault data with equipment operating parameters (such as usage frequency and operating time) and environmental monitoring data (such as ambient temperature and humidity), correlation coefficients are calculated to identify potential fault causes. For example, the analysis might reveal that "the incidence of heater overheat protection faults increases by 50% when the ambient temperature exceeds 30°C." These analytical results form a statistical report on the distribution of fault types, providing a basis for fault prevention and product improvement.

[0211] Step S6.3: Based on the statistical distribution of the fault types, combined with the equipment runtime and usage environment parameters, analyze the fault timing characteristics and environmental correlation to obtain a fault mode analysis report.

[0212] Based on the statistical distribution of fault types, further analysis is conducted on the temporal characteristics and environmental relevance of faults. For example, it analyzes whether a certain type of fault is significantly correlated with the equipment's usage time (e.g., the failure rate increases significantly after more than 12 months of use) or whether it is related to specific environmental conditions (e.g., the failure rate is higher in high-altitude areas). Through this in-depth analysis, a fault mode analysis report is obtained, revealing the underlying patterns and influencing factors of fault occurrence.

[0213] Step S6.4: Based on the fault mode analysis report, update the feature weights and threshold parameters of the fault feature library, optimize the fault diagnosis rules, and obtain an improved diagnostic model.

[0214] Based on the fault mode analysis report, the fault feature database is updated and optimized. Feature weights are adjusted to improve sensitivity to key features; threshold parameters are updated to improve the accuracy of fault diagnosis; and diagnostic rules are optimized to improve the logic of fault identification. For example, if analysis reveals that temperature sensor data is more critical for diagnosing a certain type of fault, its weight is increased; if it is found that the temperature threshold for a certain type of fault needs to be dynamically adjusted according to the ambient temperature, the corresponding threshold calculation rules are updated. Through this continuous optimization, an improved diagnostic model is obtained, enhancing the accuracy and efficiency of fault diagnosis.

[0215] Step S6.5: Integrate user feedback and maintenance records, extract new fault features and solutions, update the fault knowledge base content, and obtain the iteratively optimized fault knowledge base.

[0216] Collect and analyze user feedback on fault handling to understand user evaluations of the accuracy of fault diagnosis and the effectiveness of solutions. Simultaneously, integrate maintenance records from professional repair personnel, including information such as the actual causes of faults, handling methods, and result verification. Through analysis of this data, extract newly discovered fault characteristics and effective solutions to update the fault knowledge base. For example, if a new fault mode is discovered, it is added to the fault characteristic database; if a more effective solution is found, the corresponding handling recommendations are updated. Through this continuous knowledge accumulation and optimization, an iteratively optimized fault knowledge base is formed, providing more comprehensive and accurate support for future fault diagnosis and handling.

[0217] The following is a complete example of the troubleshooting process for intelligent capsule fresh-extract beverage machines:

[0218] A user's smart capsule beverage machine experienced insufficient water flow during beverage preparation. At this point, the machine's multi-sensor system immediately activated. The flow sensor detected a water flow rate of only 15 ml / min, far below the normal value of 50 ml / min; the pressure sensor showed a water pump outlet pressure of 1.2 bar, lower than the normal value of 2.5 bar; the current sensor detected a water pump motor current of 0.8 A, higher than the normal value of 0.5 A; and the vibration sensor detected an abnormal, irregular fluctuation in the water pump's vibration frequency. This raw sensor data was collected, transmitted to the main control unit via the internal bus, and timestamped to form a structured raw sensor data set.

[0219] The data preprocessing module processes this raw data. First, a mean-mode filtering algorithm is applied to eliminate sudden noise in the vibration data. Then, Kalman filtering is used to smooth the flow and pressure data curves. Next, the data from different types of sensors are normalized and mapped to the [0,1] interval for easier comprehensive analysis. Comparing this standardized data with a pre-established parameter benchmark dataset reveals that the flow rate deviates from the normal range by 70%, the pressure value by 52%, and the current value by 60%. Therefore, these data points are marked as anomalies, generating preprocessed sensor data with anomaly labels.

[0220] The fault diagnosis module analyzes the preprocessed data. First, the abnormal data is divided into intervals; for example, the flow rate range of 0-100 ml / min is divided into 10 intervals, resulting in an interval frequency vector [0,85,15,0,0,0,0,0,0,0], indicating that the flow rate is in the 0-10 ml / min interval 85% of the time and in the 10-20 ml / min interval 15% of the time. These interval frequency vectors are then input into a pre-trained neural network model, which contains an input layer (60 neurons), two hidden layers (128 and 64 neurons), and an output layer (32 neurons) to generate a fault feature representation. The cosine similarity of this feature representation with patterns in the fault feature library is calculated, revealing a similarity of 0.93 with the "water pump blockage" pattern, significantly higher than other fault types. Further analysis using a decision tree confirms the diagnosis as "water pump inlet pipe blockage," with the fault location at the water pump inlet, affecting the entire water flow system.

[0221] The information generation module classifies the fault based on the fault diagnosis results. The functional impact of the fault is assessed at 4 points (moderate impact), the safety risk level at 1 point (no safety risk), and the repair difficulty at 2 points (user can handle it themselves), for a total score of 7, classifying it as a "Level B fault". A corresponding template is extracted from the fault description template library, specific parameters are filled in, and a fault description is generated: "Insufficient water flow detected. Current flow rate is 15ml / min, lower than the normal value of 50ml / min." Based on historical data analysis, a cause analysis is generated: "Possible causes: 1. Blockage in the water pump inlet pipe (75%); 2. Water tank lack of water (15%); 3. Water pump malfunction (10%)." For the most likely cause, a solution is generated: "1. Turn off the power; 2. Remove the water tank; 3. Check and clean the inlet filter at the bottom of the water tank; 4. Reinstall the water tank and ensure it is correctly positioned; 5. Restart the equipment and try again." This information is integrated into structured fault information.

[0222] The information output module classifies fault information into fault warning messages (Level B) and configures a yellow background display scheme. The device's local display shows "Fault: Insufficient water flow [Level B]" and a brief handling prompt "Please check the water tank inlet," while a yellow LED indicator flashes slowly. Detailed fault information is also transmitted to the user's mobile app via the MQTT protocol. The app displays the fault diagnosis process, three possible causes and their probabilities, and a detailed five-step solution with illustrated instructions. Simultaneously, the fault diagnosis results are converted to JSON format, encrypted, and uploaded to the cloud server.

[0223] The cloud server receives and decrypts this fault data, adding it to a structured fault data table. It was found that 27 devices of the same model recently reported similar faults. Through hierarchical differential processing to calculate the fault frequency, it was discovered that this fault occurred more frequently in devices used for 3-6 months and was positively correlated with water hardness. A fault frequency heatmap was generated, showing that the fault rate in hard water areas was 40% higher than in soft water areas. Based on this analysis, the fault feature database was updated, increasing the weight of water quality factors in fault diagnosis, and a new prevention recommendation was added to the knowledge base: "Users in hard water areas are advised to clean the water tank inlet filter once a month." This update will be applied to subsequent fault diagnosis to improve fault prevention and handling efficiency.

[0224] like Figure 6 As shown, the present invention also provides a fault information processing system for an intelligent capsule fresh-extract beverage machine, comprising:

[0225] The data acquisition module is used to collect real-time data on temperature, pressure, flow rate, current, voltage and vibration frequency through multiple sensors deployed in key components of the intelligent capsule fresh-extract beverage machine, and obtain raw sensor data.

[0226] The data preprocessing module is used to filter and normalize the raw sensor data to obtain preprocessed sensor data.

[0227] The fault diagnosis module is used to perform multidimensional fault feature analysis and matching based on the preprocessed sensor data, using a polynomial-time approximation algorithm of complete interval degree, to obtain fault diagnosis results.

[0228] The information generation module is used to classify the faults according to their severity and user manageability based on the fault diagnosis results, and generate structured fault information that includes fault description, cause analysis and solutions.

[0229] The information output module is used to output the structured fault information through the local display interface of the device and the mobile terminal, and to upload the fault diagnosis results to the cloud server.

[0230] The data processing module is used to process the fault diagnosis results and update the fault knowledge base by applying a top-down hierarchical differential private counting query mechanism in the cloud server.

[0231] This invention achieves accurate diagnosis, graded processing, and multi-channel output of faults in intelligent capsule fresh-extract beverage machines through multi-sensor real-time data acquisition, data preprocessing and anomaly detection, multi-dimensional fault diagnosis and location, hierarchical fault information generation, multi-channel fault information output, fault data storage and analysis, and fault handling feedback and knowledge base updates. This improves fault handling efficiency and enhances user experience.

[0232] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0233] It should be noted that those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of this invention. If such modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include such modifications and variations.

[0234] This disclosure also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program performs the steps of a fault information processing method for an intelligent capsule fresh-extract beverage machine as described in the above-described method embodiments. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0235] Furthermore, this disclosure also provides a computer program product storing a computer program. When the computer program is run by a processor, it executes the steps of any of the above-described intelligent capsule fresh-extract beverage machine fault information processing methods. For details, please refer to the above-described method embodiments, which will not be repeated here.

[0236] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium, which can be a volatile or non-volatile computer-readable storage medium. In another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0237] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and apparatuses described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0238] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0239] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0240] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0241] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.

Claims

1. A method for handling fault information in an intelligent capsule fresh-extract beverage machine, characterized in that, include: By deploying multiple sensors on key components of the intelligent capsule fresh-extract beverage machine, real-time data on temperature, pressure, flow rate, current, voltage, and vibration frequency are collected to obtain raw sensor data. The original sensor data is filtered and normalized to obtain preprocessed sensor data. Based on the preprocessed sensor data, a multidimensional fault feature analysis and matching are performed using a polynomial-time approximation algorithm of complete interval degree to obtain fault diagnosis results. Based on the fault diagnosis results, the faults are classified according to their severity and user manageability, and structured fault information containing fault descriptions, cause analysis, and solutions is generated. The structured fault information is output through the device's local display interface and mobile terminal, and the fault diagnosis results are uploaded to the cloud server; The fault diagnosis results are processed and the fault knowledge base is updated in the cloud server using a top-down hierarchical differential private counting query mechanism. Based on the preprocessed sensor data, a multidimensional fault feature analysis and matching are performed using a polynomial-time approximation algorithm of complete interval degree to obtain fault diagnosis results, including: Collect and organize historical fault cases, extract fault characteristic indicators and corresponding sensor data characteristics, and obtain a fault characteristic mapping table; Based on the fault feature mapping table, a fault feature library containing fault type, sensor data features and occurrence scenario is established to obtain fault feature reference data. The preprocessed sensor data with anomaly markers is divided into intervals, and the data distribution density of each interval is calculated to obtain the interval frequency vector. The similarity between the interval frequency vector and the fault feature reference data is calculated to obtain a fault type candidate set; Component association analysis is performed on the candidate set of fault types to determine the location and scope of the fault and obtain the fault diagnosis results.

2. The method according to claim 1, characterized in that, By deploying multiple sensors on key components of the intelligent capsule fresh-extract beverage machine, real-time data on temperature, pressure, flow rate, current, voltage, and vibration frequency are collected to obtain raw sensor data, including: Based on the preset sampling time interval and sampling trigger conditions, the analog signals of each sensor are sampled at regular intervals to obtain time-series sampling data; The time-series sampling data is subjected to analog-to-digital conversion to obtain digital sensor data; The digital sensor data is transmitted to the main control unit via an internal bus, and a timestamp is added to the received digital sensor data in the main control unit to obtain timestamped sensor data. A data cache is established in the main control unit, and the timestamp-marked sensor data is indexed according to sensor type and acquisition time to obtain a structured raw sensor data set.

3. The method according to claim 2, characterized in that, The raw sensor data is preprocessed by filtering and normalization to obtain preprocessed sensor data, including: The structured raw sensor data set is subjected to mean filtering and Kalman filtering algorithms to eliminate noise interference and obtain filtered data. The filtered data is normalized to map different types of data to a unified interval, resulting in standardized data. Based on statistical analysis of historical operational data, normal value ranges and fluctuation thresholds for data from each sensor are established to obtain a parameter benchmark dataset. The standardized data is compared with the parameter benchmark dataset, and abnormal data points are marked according to a preset deviation threshold to obtain preprocessed sensor data with abnormal markings.

4. The method according to claim 1, characterized in that, The preprocessed sensor data with anomaly markers is divided into intervals, and the data distribution density of each interval is calculated to obtain the interval exponent vector, including: The range of values ​​for each sensor data is divided into equal intervals, and a numerical interval mapping table is established to obtain the interval division scheme. Based on the interval division scheme, the data distribution frequency of each interval is counted to generate the interval frequency vector; The similarity between the interval frequency vector and the fault feature reference data is calculated to obtain a fault type candidate set, including: A multi-layer neural network model is constructed to extract features from the interval frequency vector to obtain fault feature representation; The cosine similarity between the fault feature representation and the preset pattern in the fault feature reference data is calculated to obtain a similarity matrix; Based on the similarity matrix, fault types with similarity exceeding a preset threshold are filtered to obtain preliminary fault types; By combining the initial fault assessment results from multiple sensors, a decision tree is constructed to determine the fault type, resulting in a candidate set of fault types.

5. The method according to claim 1, characterized in that, Based on the fault diagnosis results, the faults are classified according to severity and user manageability, generating structured fault information that includes fault description, cause analysis, and solutions, including: Based on the degree of impact on equipment functionality, the level of safety risk, and the difficulty of maintenance, a fault assessment index system is established to obtain the assessment benchmark. Based on the aforementioned evaluation criteria, the fault diagnosis results are quantitatively scored, fault levels are classified, and fault classification results are obtained. Extract the description template corresponding to the fault classification result from the preset fault description template library, fill in the fault parameters and the scope of influence, and obtain the fault description content; Based on statistical analysis of historical fault data, the probability distribution of different fault causes is calculated, and cause analysis content is generated. Based on the fault classification results, select the corresponding handling plan, and combine the equipment status generation inspection, operation and verification steps to obtain the solution content; The fault description, cause analysis, and solution content are integrated into a unified format to obtain structured fault information.

6. The method according to claim 1, characterized in that, The structured fault information is output through the device's local display interface and mobile terminal, and the fault diagnosis results are uploaded to the cloud server, including: The structured fault information is divided into emergency alarm information, fault prompt information, and maintenance suggestion information according to the fault level, resulting in a hierarchical information set; Based on the hierarchical information set, the basic information layout of the local display interface and the detailed information layout of the mobile terminal are configured to obtain a display configuration scheme. According to the display configuration scheme, the fault name, alarm level and basic processing prompts are output on the local display interface of the device, and the alarm level is output through different colors and flashing frequencies of LED indicator lights; The structured fault information is transmitted to the user's mobile terminal via the MQTT protocol. The fault diagnosis process, cause analysis, and handling solutions are displayed in layers according to the importance of the information, and the mobile terminal output information is obtained. The fault diagnosis results are encapsulated in JSON format and encrypted with AES, and then uploaded to the cloud server via WiFi or Bluetooth communication.

7. The method according to claim 1, characterized in that, The fault diagnosis results are processed using a top-down hierarchical differential private counting query mechanism on the cloud server to update the fault knowledge base, including: The cloud server receives and decrypts the fault diagnosis results uploaded by multiple devices, establishes a structured fault data table, and obtains the original fault statistics data. The original fault statistics data are subjected to hierarchical differential processing to calculate the frequency of fault occurrence while protecting user privacy, and to obtain fault type distribution statistics. Based on the statistical distribution of the aforementioned fault types, combined with equipment runtime and usage environment parameters, the timing characteristics and environmental correlation of faults are analyzed to obtain a fault mode analysis report. Based on the fault mode analysis report, update the feature weights and threshold parameters of the fault feature library, optimize the fault diagnosis rules, and obtain an improved diagnostic model. By integrating user feedback and maintenance records, extracting new fault characteristics and solutions, and updating the fault knowledge base, an iteratively optimized fault knowledge base is obtained.

8. The method according to claim 7, characterized in that, The original fault statistics are subjected to hierarchical differential processing to calculate the frequency of fault occurrence while protecting user privacy, resulting in fault type distribution statistics, including: The original fault statistics data are indexed in multiple layers according to equipment model, usage time, geographical region and ambient temperature to obtain a hierarchical data structure. In each layer of data index, fault records are grouped and counted, and Laplace noise is added according to the differential privacy budget to obtain the privatized counting result; Based on the privatized counting results, the time distribution characteristics of the fault occurrence frequency are calculated using a sliding time window to obtain fault trend data. Hierarchical clustering analysis is performed on the fault trend data to generate a fault frequency heatmap and a time-series trend map, resulting in visual statistical charts; By combining equipment operating parameters and environmental monitoring data, correlation analysis is used to identify the causes of failures and obtain a statistical report on the distribution of failure types.

9. A fault information processing system for an intelligent capsule fresh-extract beverage machine, characterized in that, include: The data acquisition module is used to collect real-time data on temperature, pressure, flow rate, current, voltage and vibration frequency through multiple sensors deployed in key components of the intelligent capsule fresh-extract beverage machine, and obtain raw sensor data. The data preprocessing module is used to filter and normalize the raw sensor data to obtain preprocessed sensor data. The fault diagnosis module is used to perform multidimensional fault feature analysis and matching based on the preprocessed sensor data, using a polynomial-time approximation algorithm of complete interval degree, to obtain fault diagnosis results. The information generation module is used to classify the faults according to their severity and user manageability based on the fault diagnosis results, and generate structured fault information that includes fault description, cause analysis and solutions. The information output module is used to output the structured fault information through the local display interface of the device and the mobile terminal, and to upload the fault diagnosis results to the cloud server. The data processing module is used to process the fault diagnosis results and update the fault knowledge base by applying a top-down hierarchical differential private counting query mechanism in the cloud server. Based on the preprocessed sensor data, a multidimensional fault feature analysis and matching are performed using a polynomial-time approximation algorithm of complete interval degree to obtain fault diagnosis results, including: Collect and organize historical fault cases, extract fault characteristic indicators and corresponding sensor data characteristics, and obtain a fault characteristic mapping table; Based on the fault feature mapping table, a fault feature library containing fault type, sensor data features and occurrence scenario is established to obtain fault feature reference data. The preprocessed sensor data with anomaly markers is divided into intervals, and the data distribution density of each interval is calculated to obtain the interval frequency vector. The similarity between the interval frequency vector and the fault feature reference data is calculated to obtain a fault type candidate set; Component association analysis is performed on the candidate set of fault types to determine the location and scope of the fault and obtain the fault diagnosis results.

Images