A monitoring method and system of a cooking apparatus

Abstract

The application discloses a kind of monitoring method and system of cooking equipment, the method includes: obtaining the real-time multi-source heterogeneous sensor data of cooking equipment;Physical quantity in multi-source heterogeneous sensor data is mapped as the motion parameter of mechanical component, and motion state vector is obtained;According to the motion state vector, the pre-constructed three-dimensional visualization model is driven to realize the real-time synchronous mapping of cooking equipment operating state and virtual model;Based on multi-source heterogeneous sensor data and motion state vector, the operating state of each component is evaluated, and component-level operation evaluation result is obtained;According to the evaluation result, render visualization state identifier on the component corresponding to model, and show component-level operating state.The application solves the problem that equipment monitoring is not intuitive, multi-source heterogeneous sensor data and mechanical action lack semantic correlation and fault positioning is slow, and realizes the effect of intuitive three-dimensional monitoring, unified adaptation and rapid fault positioning.

Description

Technical Field

[0001] This invention relates to the field of intelligent cooking equipment technology, and in particular to a method and system for monitoring cooking equipment. Background Technology

[0002] With the intelligent transformation of the catering industry, commercial intelligent cooking robots have been widely used in chain restaurants and unmanned restaurants. These devices integrate complex electromechanical systems such as electromagnetic heating, multi-degree-of-freedom stirring, automatic feeding, and multi-type seasoning delivery, enabling them to automatically complete the entire cooking process without human intervention. The development of digital twin and 3D visualization technologies has provided new technical means for the remote monitoring of complex equipment.

[0003] Currently, monitoring solutions for intelligent cooking robots typically employ built-in data acquisition modules. These modules transmit sensor data, such as core parameters, motor current, proximity switch status, and pot temperature, to the cloud or local server via protocols like MQTT. The data is then presented to maintenance personnel in the form of two-dimensional charts or numerical lists. When an anomaly occurs, the system triggers an alarm based on a threshold, indicating that a certain parameter has exceeded its limit. Maintenance personnel must then remotely assess the situation by reviewing alarm records and device logs. If necessary, they should contact on-site personnel to take photos or describe the fault symptoms for further investigation.

[0004] In the above monitoring scheme, maintenance personnel need to have professional electromechanical knowledge to judge the status of equipment from abstract data, which has a high cognitive threshold and low judgment efficiency; remote personnel cannot intuitively know the specific physical location of the fault, and relying on on-site descriptions leads to high communication costs and slow positioning; moreover, different models have large differences in mechanical structure, lack a unified visualization adaptation framework, and lack semantic correlation between the underlying sensor data and macroscopic mechanical movement, making it difficult to achieve data-driven virtual-real synchronization. Summary of the Invention

[0005] This invention provides a method and system for monitoring cooking equipment, which solves the problems of high judgment threshold caused by the reliance on abstract data and two-dimensional charts in monitoring cooking equipment, lack of semantic correlation between sensor data and mechanical actions, and the inability to quickly locate faulty parts due to ambiguous operating status. It achieves the technical effect of reducing the cognitive threshold of monitoring, bridging the semantic gap, and quickly locating faults by mapping physical quantities in multi-source heterogeneous sensor data to motion parameters of mechanical parts to drive real-time synchronous mapping of a three-dimensional visualization model, and intuitively rendering status indicators on the three-dimensional model in conjunction with component-level operating status assessment.

[0006] This invention provides a method for monitoring cooking equipment, the method comprising: Acquire multi-source heterogeneous sensor data of the cooking device during operation, wherein the multi-source heterogeneous sensor data includes electrical parameters, moving part status parameters, and temperature parameters; The physical quantities in the multi-source heterogeneous sensor data are mapped to the motion parameters of the mechanical components to obtain the motion state vector; The pre-built 3D visualization model is driven by the motion state vector to achieve real-time synchronous mapping between the operating status of the cooking equipment and the virtual model. Based on the multi-source heterogeneous sensor data and historical operating data, the operating status of each component of the cooking equipment is evaluated to obtain component-level operating status evaluation results. Based on the operational status assessment results, visual status markers are rendered on the corresponding components of the 3D visualization model to display the component-level operational status.

[0007] Optionally, the step of mapping the physical quantities in the multi-source heterogeneous sensor data to motion parameters of the mechanical component to obtain the motion state vector includes: The hierarchical semantic transformation is performed on the multi-source heterogeneous sensor data. The hierarchical semantic transformation includes signal-level transformation that maps the original physical quantities in the multi-source heterogeneous sensor data to standardized physical quantities, component-level transformation that maps the standardized physical quantities to the motion state of mechanical components, and action-level transformation that infers macroscopic cooking actions by combining the operating context of the cooking equipment. Based on the result of the hierarchical semantic transformation, a motion state vector is generated.

[0008] Optionally, the action-level transition includes: Determine the recipe step currently being executed by the cooking device and the corresponding context information for that recipe step; Based on the context information of the recipe process and the multi-source heterogeneous sensor data, the type of cooking action currently being performed is determined according to a preset mapping relationship; The cooking action type is written as an attribute parameter into the motion state vector.

[0009] Optionally, the step of mapping standardized physical quantities to component-level transformations of the motion states of mechanical components includes: Obtain the motor current from the standardized physical quantities, as well as the no-load current, rated current, rated speed, and slip ratio of the motor under no-load conditions; When the motor current is greater than a set multiple of the no-load current, the load rate is calculated based on the ratio of the difference between the motor current and the no-load current to the difference between the rated current and the no-load current. The current speed is estimated based on the load rate, the rated speed, and the slip ratio.

[0010] Optionally, the method further includes: Call the pre-built parametric model library to obtain a superset skeletal tree covering the motion degrees of freedom of all models; Based on the configuration file corresponding to the model identifier of the cooking equipment, the bone nodes in the superset skeleton tree are enabled / disabled and their geometric parameters are adjusted. Based on the adjusted skeletal tree, the three-dimensional visualization model corresponding to the cooking equipment is dynamically assembled and generated.

[0011] Optionally, the step of driving the pre-built 3D visualization model based on the motion state vector includes: Obtain the arrival time interval of the sensor data frames corresponding to the motion state vector; The target animation driving mode is determined based on the time interval, and the corresponding target state parameters are calculated. The target state parameters are applied to the skeletal nodes of the pre-built 3D visualization model to update the local skeletal matrix of the 3D visualization model. The updated skeletal matrix is ​​used to render and output a 3D visualization, which drives the virtual model components to move synchronously, thus achieving real-time synchronous mapping between the operating status of the cooking equipment and the virtual model.

[0012] Optionally, the step of determining the target animation driving mode based on the time interval and calculating the corresponding target state parameters includes: If the time interval is less than or equal to the first threshold, the target state parameters of the three-dimensional visualization model are calculated, starting from the current state of each component of the pre-constructed three-dimensional visualization model and ending with the target state in the current motion state vector. If the time interval is greater than a first threshold and less than a second threshold, then multiple target motion state vectors are determined, and the target state parameters of the three-dimensional visualization model are calculated based on the multiple target motion state vectors; wherein, the multiple motion state vectors include the current motion state vector and multiple historical motion state vectors that are temporally adjacent. If the time interval is greater than or equal to the second threshold, the current state parameter of the 3D visualization model is determined as the target state parameter, and a visual cue parameter is generated; wherein, the visual cue parameter is used to indicate the display of a visual cue of data interruption on the 3D visualization model.

[0013] Optionally, the step of evaluating the operational status of each component of the cooking equipment based on the multi-source heterogeneous sensor data and the motion state vector to obtain component-level operational status evaluation results includes: The subsystems to be evaluated in the cooking equipment are determined based on the model identifier. For each subsystem to be evaluated, a weighted calculation is performed based on the multi-source heterogeneous sensor data, the motion state vector, and historical statistical data to obtain the subsystem's operational score; The operational scores of each subsystem are used as the results of the component-level operational status assessment.

[0014] Optionally, the method further includes the step of acquiring real-time multi-source heterogeneous sensor data of the cooking equipment: Receive historical sensor data from the cooking device and put the historical sensor data into a memory blocking queue buffer; Multiple consumer threads consume data from the blocking queue in parallel and write it to the time-series database in batches. The latest data frame in the time series database is identified as the real-time multi-source heterogeneous sensor data.

[0015] Optionally, the method further includes: Establish a keyframe index for historical time-series data; In response to playback control commands, the nearest keyframe is located using a binary search algorithm; The precise state of the target time is calculated based on the most recent keyframe and interpolation algorithm, and then the 3D visualization model is used to display it.

[0016] Furthermore, to achieve the above objectives, embodiments of the present invention also provide a monitoring system for cooking equipment, the system comprising: The data acquisition module is used to acquire multi-source heterogeneous sensor data of the cooking equipment in operation, wherein the multi-source heterogeneous sensor data includes electrical parameters, moving part status parameters and temperature parameters; The semantic conversion module is used to map the physical quantities in the multi-source heterogeneous sensor data into motion parameters of mechanical components to obtain a motion state vector. The motion drive module is used to drive a pre-built three-dimensional visualization model according to the motion state vector, so as to realize the real-time synchronous mapping between the operating status of the cooking equipment and the virtual model. The operation evaluation module is used to evaluate the operation status of each component of the cooking equipment based on the multi-source heterogeneous sensor data and historical operation data, and obtain the component-level operation status evaluation results. The visualization module is used to render visualization status markers on the corresponding components of the three-dimensional visualization model based on the operational status evaluation results, so as to display the operational status at the component level.

[0017] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages: By mapping the physical quantities in the multi-source heterogeneous sensor data collected during the operation of cooking equipment to the motion parameters of mechanical components, and driving the real-time synchronous mapping of the 3D visualization model, maintenance personnel can intuitively understand the equipment's operating status without professional knowledge, thus lowering the cognitive threshold for monitoring. At the same time, based on the multi-source heterogeneous sensor data and historical operating data, the operating status of each component is evaluated at the component level, and the corresponding component is rendered with a visual status label in the 3D visualization model, realizing intuitive location and rapid identification of faulty parts, effectively improving the efficiency of remote operation and maintenance. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the framework of the monitoring method for cooking equipment according to Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the monitoring system for the cooking equipment according to Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the terminal structure of the hardware operating environment involved in Embodiment 3 of the present invention. Detailed Implementation

[0019] To address the problems of relying on abstract data and two-dimensional charts in cooking equipment monitoring, which makes it difficult for maintenance personnel to intuitively understand equipment status, hinders unified adaptation across different models, lacks semantic correlation between sensor data and mechanical actions, and results in ambiguous descriptions of equipment operating status that prevent rapid location of faulty components, this invention proposes a monitoring method and system for cooking equipment. By acquiring real-time multi-source heterogeneous sensor data and model identification, a corresponding 3D visualization model is parameterized and loaded. Layered semantic transformation is used to map physical quantities into motion state vectors and infer cooking actions. Adaptive interpolation and prediction algorithms drive the real-time synchronous movement of model components and the physical equipment. Simultaneously, component-level operational evaluations of each subsystem are performed based on multi-source heterogeneous sensor data and motion state vectors. Finally, color-coded labels and fault indications are rendered on the corresponding components in the 3D model. This solution can transform discrete, low-frequency, multi-source heterogeneous sensor data into continuous, intuitive animations, adapting to various models, bridging the semantic gap between data and action, and presenting component operating status in a visual manner. This lowers the monitoring threshold and improves remote maintenance efficiency and fault location speed.

[0020] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the invention to those skilled in the art.

[0021] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0022] Example 1 In this embodiment, a method for monitoring cooking equipment is provided.

[0023] Reference Figure 1 The monitoring method for the cooking equipment in this embodiment includes the following steps: Step S1: Acquire multi-source heterogeneous sensor data of the cooking device in operation, wherein the multi-source heterogeneous sensor data includes electrical parameters, moving part status parameters and temperature parameters; In this embodiment, electrical parameters include input voltage, input current, output power, operating frequency, and IGBT temperature. Moving part status parameters include spindle motor current, powder motor current, sauce motor current, and proximity switch status indicator; temperature parameters include the pot's internal temperature.

[0024] As an optional implementation, when acquiring real-time multi-source heterogeneous sensor data, the system receives multi-source heterogeneous sensor data from the cooking device and buffers it in a memory blocking queue. Multiple consumer threads consume the data in parallel from the blocking queue and write it to a time-series database in batches. The latest data frame is then read from the time-series database as the real-time multi-source heterogeneous sensor data. By using a memory queue buffer and batch writing, the system can handle high-throughput data reporting, reduces data processing latency through parallel consumption, and ensures data persistence and efficient querying through the time-series database.

[0025] For example, the collected multi-source heterogeneous sensor data is serialized in Protobuf format using the MQTT protocol. This data includes 27 core parameters, 18 motor currents, 4 sets of proximity switch states, and the pot's internal temperature signal. The serialized data is parsed to extract data points for each sampling time. Each data point contains a complete sensor value and its corresponding timestamp. Then, each data point is converted into a unified data structure and placed into a bounded blocking queue with a capacity of 100,000. The MQTT processing thread acts as the producer, writing to the queue in a non-blocking manner. When the queue is full, the oldest data is discarded and an alarm is recorded. Additionally, three consumer threads are started, each triggering a batch of 5000 records or a 5-second timeout, to write to the time-series database in batches via the TDengine Schemaless REST API. An exponential backoff retry strategy is used to handle write failures, with the retry interval increasing progressively. Retrying stops after reaching a preset maximum number of retries, and a failure log is recorded. For example, the retry intervals are 1 second, 2 seconds, and 4 seconds, with a total of 3 retries. Simultaneously, the latest data frame is read from the time-series database and incrementally pushed to the client via WebSocket for subsequent steps.

[0026] Optionally, the multi-source heterogeneous sensor data also includes vibration sensor data, oil fume concentration sensor data, and food weight multi-source heterogeneous sensor data.

[0027] Optionally, after acquiring multi-source heterogeneous sensor data, the model identifier can also be acquired, and the corresponding 3D visualization model can be loaded based on the model identifier; In this embodiment, the model identifier is a unique identifier for the equipment, such as 100EA or 300ES. The 3D visualization model refers to a digital twin of the cooking equipment, capable of displaying the equipment's shape, structure, and moving parts in three-dimensional space. Different models differ in pot size, number of stirring arms, feeding mechanism, and seasoning pipeline layout. This step generates a model matching the model through parametric methods, eliminating the need to create a complete model for each model; only configuration files and differentiated Mesh resources are required.

[0028] As an optional implementation, when loading the 3D visualization model, a pre-built parametric model library is invoked to obtain a superset skeleton tree covering the motion degrees of freedom of all models. Based on the configuration file corresponding to the model identifier, the bone nodes in the superset skeleton tree are enabled / disabled and their geometric parameters are adjusted. Based on the adjusted skeleton tree and corresponding mesh resources, the 3D visualization model is dynamically assembled and generated. By using a unified skeleton tree and configuration file, a single rendering logic can be adapted to all models. Adding a new model only requires adding a configuration file and differentiated mesh resources, significantly reducing modeling and maintenance costs.

[0029] For example, a superset skeletal tree is predefined, with a corresponding JSON configuration file for each machine model. This configuration file describes which nodes in the superset skeletal tree are enabled for that model, the geometric parameters and motion constraints of each node, and the Mesh resource path corresponding to the node's appearance. For example, for the "100EA" automatic feeding machine model, its configuration file can specify: pot body enabled, capacity 20L; tilt joint tilt range [-5°, 120°]; stirring arm enabled, quantity 1, length 350mm, speed range 0~120rpm; feeding unit enabled, material box turntable 8 slots; liquid seasoning unit enabled, liquid pump quantity 6, layout front; powder seasoning unit enabled, powder motor quantity 4, layout top; sauce unit enabled, sauce motor quantity 2, layout together with powder; heating power 22 kW, voltage specification 380V, etc. For the manual feeding machine model, its configuration file sets the feeding unit and related nodes to disabled.

[0030] Load the corresponding configuration file based on the current device model identifier, and traverse all nodes in the superset skeletal tree: for nodes marked as enabled in the configuration file, set the node's size, position offset, and other attributes according to the geometric parameters in the configuration file, set the node's motion range, such as joint limits and speed limits, according to motion constraints, and record the Mesh resource path corresponding to the node; for nodes marked as disabled in the configuration file, directly mark the node and all its child nodes as invisible and do not participate in subsequent animation updates.

[0031] After completing the above configuration, the actual 3D model file is loaded according to the Mesh resource path of each node, and these Mesh resources are bound to the corresponding skeletal nodes to form a complete skeletal animation model. For enabled nodes, their Mesh resources will change with the movement of the skeletal nodes; for disabled nodes, no Mesh resources are loaded and they do not participate in rendering. Finally, a complete 3D digital twin instance of the device is generated.

[0032] Optionally, the superset skeletal tree includes the following skeletal nodes: Root (device base), Body (body frame), HeatingUnit (heating unit), Pot (pot body), PotTilt (pot body tilting joint), StirringArm (stirring arm, expandable to multiple), FeedingUnit (feeding unit), FeedingArm (feeding arm), MaterialTray (material tray turntable), LiquidCondimentUnit (liquid condiment unit), LiquidPump (liquid pump, expandable to multiple), PowderCondimentUnit (powder condiment unit), PowderMotor (powder screw motor, expandable to multiple), PasteCondimentUnit (sauce unit), PasteMotor (sauce plunger motor, expandable to multiple), ControlPanel (control panel), Base (base), etc. This superset skeletal tree is an abstract logical skeleton, without specific geometric dimensions and appearance, only defining the hierarchical relationship between nodes and the possible motion types of each node, such as rotation and translation.

[0033] In this implementation, through the aforementioned parameterized configuration and dynamic assembly method, a single rendering logic can be adapted to all device models. When a new device model is needed, only a new configuration file and differentiated Mesh resources, such as pots or stirring arms of different shapes, need to be added. There is no need to modify the core code of the rendering engine, nor is it necessary to create a complete model for each device model, thereby significantly reducing modeling costs and system maintenance complexity.

[0034] Step S2: Map the physical quantities in the multi-source heterogeneous sensor data to the motion parameters of the mechanical components to obtain the motion state vector; In this embodiment, the motion state vector is structured data containing the target angle or displacement, motion speed, and action type label of each movable skeleton node in the skeletal tree, which can be directly used to drive the 3D model. Movable skeleton nodes include, for example, the tilting joint of the pot body and the rotating joint of the stirring arm. This embodiment converts the physical quantities of multi-source heterogeneous sensor data into motion parameters of mechanical components through hierarchical semantics. That is, it transforms the abstract data such as current, voltage, and switching quantities at the bottom layer into understandable motion parameters, such as "the stirring arm is rotating at 60 rpm" and "the pot body is tilted at 30 degrees," to bridge the semantic gap between multi-source heterogeneous sensor data and mechanical actions, thus obtaining the motion state vector.

[0035] As an optional implementation, the above-mentioned step of mapping physical quantities in multi-source heterogeneous sensor data to motion parameters of mechanical components to obtain motion state vectors includes: performing hierarchical semantic transformation on the multi-source heterogeneous sensor data, which includes signal-level transformation of mapping the original physical quantities in the multi-source heterogeneous sensor data to standardized physical quantities, component-level transformation of mapping standardized physical quantities to motion states of mechanical components, and action-level transformation of inferring macroscopic cooking actions in conjunction with the operating context of the cooking equipment; and generating motion state vectors based on the results of hierarchical semantic transformation, which include motion parameters of mechanical components and corresponding cooking action types.

[0036] Specifically, raw physical quantities refer to the values ​​directly collected by the sensor, such as the AD sampling value of voltage, the raw reading of current, and the bit combination of proximity switch. These values ​​may have different dimensions and have not been normalized.

[0037] Standardized physical quantities refer to values ​​after unit conversion and dimension unification, such as converting voltage to volts, current to amperes, power to kilowatts, temperature to degrees Celsius, and switch states to logic true / false.

[0038] The motion state of mechanical components refers to the actual motion parameters of the components after calculation, such as the tilt angle of the pot (degrees), the rotational speed of the stirring arm (revolutions per minute), the current slot number of the feeding arm, the on / off status of the seasoning pump, and the estimated flow rate (ml / s). For example, when estimating the rotational speed of the stirring arm, the motor current in the standardized physical quantities is first obtained, along with the no-load current, rated current, rated speed, and slip ratio of the motor under no-load conditions. When the motor current is detected to be greater than a set multiple of the no-load current, the load rate is calculated based on the ratio of the difference between the motor current and the no-load current to the difference between the rated current and the no-load current. Finally, the current rotational speed is estimated based on the load rate, rated speed, and slip ratio. The tilt angle of the pot can be inferred from the combination state of the proximity switch and the pot motor current.

[0039] The runtime context refers to the recipe workflow context, which includes the currently executing recipe step, the device's automatic / manual operating mode, and the historical state sequence.

[0040] Optionally, when estimating the stirring arm speed in the component-level conversion, the motor current in the standardized physical quantities, as well as the no-load current, rated current, rated speed and slip ratio of the motor under no-load conditions are obtained; when the motor current is greater than a set multiple of the no-load current, the load rate is calculated based on the ratio of the difference between the motor current and the no-load current to the difference between the rated current and the no-load current; the current speed is estimated based on the load rate, rated speed and slip ratio.

[0041] For example, if the multiplier is set to 1.1 times, the motor is determined to be in operation when the motor current exceeds 1.1 times the no-load current. The load rate is calculated as (measured current - no-load current) / (rated current - no-load current), and the current speed is calculated as rated speed × (1 - load rate × slip rate).

[0042] This implementation unifies raw data from different sources and with different dimensions into a standardized format that facilitates subsequent processing through signal-level conversion, eliminating dimensional differences. Component-level conversion associates the standardized data with specific mechanical components to obtain the motion parameters of each component; for example, it infers whether the pot is in its home, tilted, or serving position from the combination of multiple proximity switches. Action-level conversion integrates the states of multiple components and combines them with the recipe flow context to infer the cooking actions being performed by the equipment, such as preheating, stir-frying, and adding ingredients, giving the final motion state vector semantic information. Through three-level hierarchical semantic conversion, the chaotic, multi-source, heterogeneous sensor data is gradually elevated into component motion parameters and cooking action semantics with clear physical meaning, bridging the semantic gap between raw sensor data and mechanical actions. This provides accurate and easily understood input for subsequent 3D motion-driven, operational evaluation, and visualization.

[0043] As an optional implementation, when performing action-level conversion, the recipe process context currently being executed by the cooking device is obtained; based on the recipe process context and multi-source heterogeneous sensor data, the type of cooking action currently being executed is determined according to a preset mapping relationship; and the cooking action type is written as an attribute parameter into the motion state vector.

[0044] The recipe process context refers to the name of the recipe currently being executed by the cooking equipment, the current step number, the step type, and the target parameters of the step. Step types include, for example, preheating, adding ingredients, stir-frying, adding seasonings, and serving.

[0045] Target parameters can refer to target temperature, feed box number, etc.

[0046] The preset mapping relationship refers to a set of judgment rules based on multi-sensor fusion. For example, when the heating power is greater than zero and the stirring arm is not rotating and the feeding arm is not moving, it is mapped to preheating; when the feeding arm motor current exceeds the threshold and the proximity switch state changes, it is mapped to feeding, and the current material box slot position is further determined from the encoder reading; when the current of any seasoning pump rises and the estimated flow rate exceeds the threshold, it is mapped to adding seasoning, and the specific pipeline number of the rising current is used to determine which seasoning is being added, such as liquid pump No. 3; when the stirring arm motor current is continuous, the pot is in the normal working position and the heating power is greater than zero, it is mapped to stir-frying; when the pot tilt angle is greater than 90 degrees, it is mapped to serving; when none of the above conditions are met, it is mapped to standby.

[0047] This embodiment can further abstract the motion parameters of mechanical components into cooking actions that users can understand, so that the 3D twin can not only show how many degrees the pot is tilted, but also tell the user that the device is dispensing food, thus improving the intuitiveness and intelligence of monitoring.

[0048] Step S3: Drive the pre-built 3D visualization model based on the motion state vector to achieve real-time synchronous mapping between the operating status of the cooking equipment and the virtual model; In this embodiment, the sampling frequency of multi-source heterogeneous sensor data is typically low, such as 1-10Hz, while the 3D visualization model needs to be continuously rendered at a refresh rate of 60 frames per second. Directly using the latest received motion state vector to update the model position would cause noticeable jumps and stutters in the animation, severely reducing the realism of the virtual-real synchronization. To address the frequency mismatch between the low sampling rate and the high frame rate, this step employs an adaptive interpolation and prediction algorithm to convert discrete low-frequency motion state vectors into continuous, smooth high-frequency animation, while tolerating network transmission latency and brief interruptions.

[0049] As an optional implementation, when driving the 3D visualization model, the arrival time interval of the sensor data frames corresponding to the motion state vector is obtained; the target animation driving mode is determined based on the time interval, and the corresponding target state parameters are calculated; the target state parameters are applied to the skeletal nodes of the pre-built 3D visualization model to update the local skeletal matrix of the 3D visualization model; the 3D visualization screen is rendered and output based on the updated skeletal matrix to drive the virtual model components to move synchronously, thereby realizing the real-time synchronous mapping between the operating status of the cooking equipment and the virtual model.

[0050] The specific methods for determining the target animation driving mode and calculating the target state parameters include: if the time interval is less than or equal to the first threshold, the target state parameters of the 3D visualization model are calculated starting from the current state of each component of the pre-built 3D visualization model and ending at the target state in the current motion state vector; if the time interval is greater than the first threshold and less than the second threshold, multiple target motion state vectors are determined, and the target state parameters of the 3D visualization model are calculated based on the multiple target motion state vectors, including the current motion state vector and multiple historical motion state vectors that are close in time sequence; if the time interval is greater than or equal to the second threshold, the current state parameters of the 3D visualization model are determined as the target state parameters, and visual cue parameters are generated. The visual cue parameters are used to indicate the display of visual cue for data interruption on the 3D visualization model.

[0051] Specifically, when driving the model's motion, an adaptive interpolation and prediction algorithm is used to process the motion state vector. The algorithm switches between interpolation mode, prediction mode, and disconnection mode according to the time interval. In interpolation mode, an easing function is used to smoothly transition the state of the model components. In prediction mode, the motion state at the current moment is calculated by extrapolation based on historical motion trends. In disconnection mode, the last state of the model components is maintained. Based on the processed motion state, the skeleton matrix of the model components is updated to achieve motion driving.

[0052] For example, a desired data interval can be set, such as 200ms, which corresponds to a sampling frequency of 5Hz. A first threshold is then set to 300ms, and a second threshold to 3000ms. The time difference between the current moment and the moment the most recent multi-source heterogeneous sensor data frame was received is calculated. Mode switching is performed according to the following rules: If If the value is ≤ the first threshold, the data reception is considered normal, and the system enters interpolation mode; if the first threshold is < If the data is less than the second threshold, it is considered that there is a certain delay or temporary missing data, and the system enters prediction mode; if... If the value is greater than or equal to the second threshold, the data connection is considered to be interrupted, and the system enters disconnection mode.

[0053] Optionally, in interpolation mode, the motion state at the current moment smoothly transitions from the state at the previous rendering moment to the latest target state. Linear interpolation is used for linear attributes such as position and scaling; spherical linear interpolation is used for rotation attributes. The interpolation progress is set to a fixed duration, such as 250 milliseconds, and uses a smooth easing step function, whose function form is: This makes the start and end speeds of the interpolation process zero, resulting in more natural motion.

[0054] Optionally, in prediction mode, since no new multi-source heterogeneous sensor data has been received for a period of time, the new target state cannot be obtained. Therefore, the motion state at the current moment is extrapolated based on historical motion trends. That is, the motion velocity is estimated based on the received motion state vector sequence; the time difference between the current moment and the moment corresponding to the most recent multi-source heterogeneous sensor data is calculated; the predicted displacement is calculated based on the motion velocity and the time difference, and physical constraints are applied to limit the predicted displacement; the predicted displacement is superimposed on the most recent motion state to obtain the motion state at the current moment, which is used to drive the model.

[0055] Optionally, when estimating motion velocity, several recent historical motion state vectors can be saved, for example, five. Each vector contains a timestamp and the position or angle value for each degree of freedom. For each degree of freedom, a least-squares linear fit is used to obtain the rate of change of that degree of freedom over time, i.e., the motion velocity, including translational velocity or rotational angular velocity. A difference method can also be used for estimation, but linear fitting can reduce the influence of noise.

[0056] Optionally, when calculating the predicted displacement and applying physical constraints, for translational degrees of freedom, the predicted position = nearest position + translational velocity × For rotational degrees of freedom, the predicted angle = nearest angle + angular velocity × Prolonged extrapolation may cause the predicted results to exceed the physical movement range of the equipment, such as the pot tilting angle exceeding the mechanical limit. Therefore, it is necessary to limit the predicted results, that is, restrict the predicted values ​​to the movement range predefined in the configuration file, such as the tilt angle not exceeding the range of [-5°, 120°]. At the same time, the maximum speed can also be limited to prevent unreasonable abrupt changes.

[0057] Optionally, in discontinuity mode, the model no longer attempts interpolation or prediction, but instead retains the last valid motion state. Simultaneously, visual cue parameters are generated to instruct the 3D visualization model to display visual cues indicating data interruption. For example, the overall transparency of the model could be periodically varied between semi-transparent and normal, or text labels indicating data interruption could be displayed around the model.

[0058] Optionally, in each rendering frame, based on the final determined motion state at the current moment (which could be the interpolated state, the predicted state, or the maintained state), the local transformation matrix of each skeletal node in the 3D visualization model is updated, including translation, rotation, and scaling. Then, the final world coordinates of each vertex are calculated, and the graphics rendering interface is called to draw them, achieving real-time synchronization between the model parts and the actual mechanical movements of the cooking equipment.

[0059] This embodiment uses adaptive switching between interpolation mode, prediction mode, and disconnection mode to convert low-frequency, discrete sensor sampling data into high-frequency, continuous 3D animation, ensuring that the movements of the 3D visualization model are highly synchronized with the actual mechanical movements of the physical cooking equipment. Even in the event of network latency or brief interruptions, the model can still maintain smooth motion through prediction mode, avoiding jumps and stutters, greatly improving the realism of the virtual-real mapping and the usability of remote monitoring.

[0060] Step S31: Render the visual effects of the cooking process in the 3D visualization model according to the cooking action type in the motion state vector.

[0061] In this embodiment, to further enhance the intuitiveness of remote monitoring, the corresponding cooking process visual effects can be rendered in real time in the three-dimensional visualization model based on the cooking action type output by the semantic conversion result.

[0062] As an optional implementation, when rendering the visual effects of the cooking process, the cooking action type in the motion state vector is first obtained. Based on the cooking action type, corresponding particle system parameters or shader parameters are triggered, and physical effects matching the cooking action type are rendered in the 3D visualization model. For example, when the cooking action type is adding seasonings, particle effects of liquid injection can be rendered at the seasoning outlet; when the cooking action type is preheating and the pot temperature is high, particle effects of oil fumes or steam can be rendered. Parameters such as particle emission speed, particle size, color, and lifespan corresponding to different cooking action types can be pre-configured according to actual needs.

[0063] This embodiment uses particle systems and shader effects to allow users to not only see the mechanical movement of the 3D model, but also to intuitively perceive the state of the ingredients, the injection of seasonings, and the dynamic effects of oil fumes and steam during the cooking process, further enhancing the realism of remote monitoring.

[0064] Step S4: Based on multi-source heterogeneous sensor data and historical operating data, evaluate the operating status of each component of the cooking equipment to obtain component-level operating status evaluation results; In this embodiment, the operational evaluation is refined to the component level, that is, the heating system, stirring system, feeding system, and seasoning system are evaluated separately. Each subsystem corresponds to one or more components in the 3D visualization model. For example, the pot body corresponds to the heating system, and the stirring arm corresponds to the stirring system. The evaluation is based on real-time multi-source heterogeneous sensor data and historical operational data, thereby deriving an operational score for each subsystem and achieving fine-grained operational diagnosis at the component level.

[0065] As an optional implementation, when evaluating the operating status, each subsystem to be evaluated is determined based on multi-source heterogeneous sensor data and historical operating data; for each subsystem to be evaluated, a weighted calculation is performed to obtain the operating score of the subsystem; and the operating score of each subsystem is used as the component-level operating status evaluation result.

[0066] Optionally, when determining the subsystems to be evaluated, the subsystems included in the cooking equipment can be identified based on the model identifier of the cooking equipment. Different models may contain different subsystems. For example, an automatic feeding model includes a feeding system, while a manual feeding model does not. Therefore, it is necessary to query the subsystem configuration of the model based on its model identifier to determine the set of subsystems to be evaluated. For example, subsystems may include a heating system, a stirring system, a feeding system, and a seasoning system. The seasoning system can be further subdivided into a liquid pump subsystem, a powder motor subsystem, and a sauce motor subsystem.

[0067] Optionally, the motion state vector can also be used as one of the reference bases for evaluation. For example, when the motion state vector contains the actual motion parameters of the component, the actual motion parameters of the component can be compared with historical operating data to help determine whether there is any abnormality in the component.

[0068] Optionally, for each subsystem to be evaluated, a weighted calculation is performed based on real-time multi-source heterogeneous sensor data, motion state vectors, and historical statistical data, according to preset weighting coefficients, to obtain the subsystem's operational score.

[0069] For example, the operating score of the heating system can be calculated based on a weighted average of factors such as voltage deviation, power fluctuation, historical overheating frequency, and cumulative cooking frequency. A higher voltage deviation, more frequent power anomalies, more overheating frequency, or more cumulative cooking frequency results in a lower operating score. The operating score of the stirring system can be calculated based on a weighted average of factors such as the ratio of the stirring motor current to the rated current, cumulative cooking frequency, and historical stall frequency. The operating scores of the feeding and seasoning systems can be calculated based on a weighted average of factors such as the current fluctuation of their respective motors, proximity switch response delay, and cumulative usage frequency. The specific weighting coefficients for each subsystem can be preset based on practical experience. This embodiment does not limit specific weighting values ​​or calculation formulas, as long as a score reflecting the operating level of the subsystem can be calculated based on the input data.

[0070] Optionally, the operational scores of each subsystem can be organized into component-level operational status assessment results, such as a mapping table containing subsystem identifiers and corresponding operational scores. This result can be used to render operational status identifiers on the corresponding components of the 3D visualization model.

[0071] This embodiment refines the operational assessment to the component level, enabling maintenance personnel to accurately determine whether the problem lies in the heating system, stirring system, feeding system, or seasoning system, rather than just knowing a vague overall operational score. This provides a quantitative basis for highlighting faulty components on the 3D model, improving the efficiency and accuracy of fault location.

[0072] Step S5: Based on the operational status assessment results, render the visual status markers on the corresponding components of the 3D visualization model to display the component-level operational status.

[0073] In this embodiment, the component-level operational evaluation results output in step S4 are operational scores for each subsystem, such as 94 points for the heating system, 99 points for the stirring system, 95 points for the feeding system, and 98 points for the seasoning system. These scores are abstract numerical values, and maintenance personnel cannot intuitively determine which components require attention. This step converts these operational scores into intuitive visual icons, overlaying them onto the corresponding components in the 3D visualization model, enabling maintenance personnel to intuitively see which components are operating well and which require maintenance.

[0074] Optionally, when rendering the visual status indicators, the coloring parameters are determined based on the running score of each component. The following rules can be used for rendering: if the running score is ≥90, no special color is superimposed, and the original appearance of the model is maintained; if the running score is ≥70 and <90, a semi-transparent yellow halo is superimposed, indicating that the component needs attention; if the running score is <70, a semi-transparent red halo is superimposed, and a pulse flashing animation is started, indicating that the component has a serious problem that needs to be dealt with immediately; if there is a real-time fault, such as a clogging of a seasoning pump, a red highlighted border is superimposed on the corresponding component position, and a fault icon is displayed. Users can click on the icon to view the fault details.

[0075] In addition, in this embodiment, besides real-time monitoring, the historical cooking process can also be replayed in three dimensions for fault review, cooking process analysis, and operation training.

[0076] Specifically, during historical playback, a keyframe index of historical time-series data can be established; in response to playback control commands, the nearest keyframe is located using a binary search algorithm; based on the nearest keyframe and an interpolation algorithm, the precise state at the target time is calculated, and a 3D visualization model is driven for display. The historical playback function allows users to review the equipment's operating status at any historical moment, with recipe steps displayed synchronously, facilitating post-event analysis of whether the cooking process was up to standard, whether the equipment malfunctioned, and fault tracing.

[0077] For example, in the history playback function, a keyframe index of historical time-series data is first established. Each keyframe contains a timestamp and a corresponding motion state vector. When the user issues a playback control command and selects a target time, the system uses a binary search algorithm to locate the closest keyframe in the keyframe index whose timestamp is less than or equal to the target time. If the target time happens to be that keyframe, the motion state vector in that keyframe is directly used as the target state to drive the 3D visualization model. If the target time is between two keyframes, the motion state vectors of these two keyframes are interpolated to obtain the precise state of the target time, and then the 3D visualization model is driven to move, while the recipe steps corresponding to that time are displayed synchronously.

[0078] In this embodiment, by converting the hierarchical semantics of underlying multi-source heterogeneous sensor data into motion state vectors, and using these vectors to drive a 3D visualization model that is parameterized and adapted to the device model, real-time virtual-real mapping of equipment movements is achieved. This method reduces the reliance of maintenance personnel on specialized electromechanical knowledge, eliminating the need to create separate complete models for different device models, allowing for intuitive monitoring of equipment operating status. Simultaneously, adaptive interpolation and prediction algorithms ensure animation continuity and network jitter tolerance at low sampling rates, enhancing the realism of virtual-real synchronization. Furthermore, component-level operational status assessment and 3D visualization coloring make fault location fast and intuitive, significantly reducing remote communication and troubleshooting time. In addition, the historical playback function provides strong support for fault review and process analysis.

[0079] Based on the same inventive concept, the present invention also provides a system corresponding to the method in Embodiment 1, as shown in Embodiment 2.

[0080] Example 2 Based on Embodiment 1, another embodiment of the present invention is proposed, with reference to... Figure 2 The monitoring system for cooking equipment includes a data acquisition module, a semantic conversion module, a motion driving module, an operation evaluation module, and a visualization module.

[0081] The data acquisition module is used to acquire multi-source heterogeneous sensor data of the cooking equipment during operation; In this embodiment, the data acquisition module is responsible for receiving multi-source heterogeneous sensor data reported by the device, such as core parameters, motor current, proximity switch status, pot temperature, etc., and providing the pre-processed data to other modules for use.

[0082] Optionally, the data acquisition module can use message queue buffering, batch writing to a time-series database, and incremental push to ensure efficient data transmission and real-time performance.

[0083] Optionally, the system also includes a model loading module, which loads the corresponding 3D visualization model according to the model identifier; In this embodiment, the model loading module maintains a parametric model library, which pre-configures a superset skeleton tree covering the motion degrees of freedom of all device models, as well as configuration files for each model. Based on the input device model identifier, the model loading module retrieves the corresponding configuration file from the parametric model library, enables / disables the bone nodes in the superset skeleton tree, adjusts their geometric parameters, and then dynamically assembles the corresponding mesh resources to generate a complete 3D visualization model instance of the device. In this way, it is unnecessary to create a complete model separately for each device model; only configuration files and differentiated mesh resources are needed to adapt to multiple device models.

[0084] The semantic conversion module is used to map physical quantities in multi-source heterogeneous sensor data into motion parameters of mechanical components to obtain motion state vectors; In this embodiment, the semantic conversion module performs hierarchical semantic conversion on multi-source heterogeneous sensor data, including signal-level conversion, component-level conversion, and motion-level conversion. Signal-level conversion unifies the original physical quantities into standardized physical quantities; component-level conversion transforms the standardized physical quantities into the motion states of each mechanical component; and motion-level conversion infers the type of cooking action being performed based on the operating context of the cooking equipment. The final generated motion state vector contains the target angle or displacement, motion velocity, and motion type label for each movable skeleton node, which can be directly used to drive the 3D model.

[0085] The motion drive module is used to drive the movement of corresponding parts in the 3D visualization model using motion state vectors, so that the 3D visualization model and the actual mechanical movements of the cooking equipment are kept synchronized in real time. In this embodiment, the motion driving module employs an adaptive interpolation and prediction algorithm, automatically switching between interpolation mode, prediction mode, and disconnection mode based on the time interval between the most recent reception of multi-source heterogeneous sensor data. In interpolation mode, an easing function is used to smoothly transition the state; in prediction mode, physical constraints are applied based on extrapolation of historical motion trends; and in disconnection mode, the last state is maintained while providing visual cues. Then, the transformation matrix of each skeletal node in the 3D model is updated based on the processed motion state, thereby driving the model components to move continuously and smoothly.

[0086] The operation evaluation module is used to evaluate the operation status of each component of the cooking equipment based on multi-source heterogeneous sensor data and motion state vectors, and obtain component-level operation status evaluation results. In this embodiment, the operation evaluation module determines the subsystems to be evaluated within the current device based on the device model identifier. For each subsystem, a weighted calculation is performed using real-time multi-source heterogeneous sensor data, motion state vectors, and historical statistical data to obtain the operation score for that subsystem. Finally, a mapping table containing the identifiers of each subsystem and their operation scores is output, realizing component-level operation diagnosis.

[0087] The visualization module is used to render visualization status markers on the corresponding components of the 3D visualization model based on the operational status evaluation results, so as to display the operational status at the component level.

[0088] In this embodiment, the visualization module overlays different visual identifiers onto the corresponding components in the 3D model based on the operational score of each subsystem: a high operational score maintains the original appearance, a medium score is overlaid with a semi-transparent yellow halo, and a low score is overlaid with a semi-transparent red halo and starts pulse flashing; if a real-time fault exists, a red highlighted border and fault icon are additionally overlaid at the component location, and the user can click the icon to view fault details. Through this intuitive coloring method, maintenance personnel can quickly locate abnormal components without consulting data reports.

[0089] Optionally, the system also includes a physical effects rendering module, which renders the visual effects of the cooking process in a 3D visualization model based on the type of cooking action in the motion state vector.

[0090] In this embodiment, the physical effects rendering module obtains the cooking action type from the motion state vector output by the semantic conversion module, triggers the corresponding particle system parameters or shader parameters, and renders physical effects that match the cooking action type, such as food rolling, seasoning injection, or oil fumes and steam. This allows users to intuitively perceive the changes in the state of the food during the cooking process, improving the comprehensiveness and realism of the monitoring.

[0091] The modules described above work together to achieve a complete virtual-real mapping monitoring process, from data acquisition, model loading, semantic conversion, action-driven processes to operational evaluation and visualization. The specific workflow of the system in this embodiment corresponds to the method in Embodiment 1, and will not be repeated here.

[0092] It should be noted that the above modules can be software program modules running on computing devices, or functional units combining hardware and software. In actual deployment, the data acquisition module and some computing can be deployed on edge servers or cloud servers, while the visualization module runs on user terminals, such as browsers or applications on computers, tablets, or mobile phones, communicating in real time via protocols such as WebSocket. Those skilled in the art will understand that the above module division is only a logical functional division, and other splitting or merging methods may be used in actual implementation.

[0093] The technical effects of the system in this embodiment are the same as those of the method in Embodiment 1, namely, it realizes intuitive three-dimensional monitoring of cooking equipment, unified adaptation of multiple models, continuous animation under low sampling rate, component-level operation visualization, and historical playback, thereby improving the efficiency of remote operation and maintenance.

[0094] Example 3 In this embodiment of the invention, a monitoring device for cooking equipment is proposed.

[0095] Reference Figure 3 , Figure 3 This is a schematic diagram of the terminal structure of the hardware operating environment involved in an embodiment of the present invention.

[0096] like Figure 3 As shown, the control terminal may include: a processor 1001, such as a CPU, a network interface 1003, a memory 1004, and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The network interface 1003 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1004 may be high-speed RAM or stable non-volatile memory, such as disk storage. Alternatively, the memory 1004 may be a storage device independent of the aforementioned processor 1001.

[0097] Those skilled in the art will understand that Figure 3 The terminal structure shown does not constitute a limitation on the terminal and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0098] like Figure 3 As shown, the memory 1004, which serves as a computer storage medium, may include an operating system, a network communication module, and a monitoring program for the cooking equipment.

[0099] exist Figure 3 In the hardware structure of the monitoring device for the cooking equipment shown, the processor 1001 can call the monitoring program for the cooking equipment stored in the memory 1004 and perform the following operations: Acquire multi-source heterogeneous sensor data of the cooking equipment during operation, including electrical parameters, moving part status parameters, and temperature parameters; By mapping the physical quantities in the data from multi-source heterogeneous sensors to the motion parameters of mechanical components, a motion state vector is obtained. A pre-built 3D visualization model is driven by motion state vectors to achieve real-time synchronous mapping between the operating status of cooking equipment and the virtual model; Based on multi-source heterogeneous sensor data and historical operating data, the operating status of each component of the cooking equipment is evaluated to obtain component-level operating status evaluation results. Based on the operational status assessment results, visual status markers are rendered on the corresponding components of the 3D visualization model to display the component-level operational status.

[0100] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Hierarchical semantic transformation is performed on multi-source heterogeneous sensor data. The hierarchical semantic transformation includes signal-level transformation that maps the raw physical quantities in the multi-source heterogeneous sensor data to standardized physical quantities, component-level transformation that maps standardized physical quantities to the motion state of mechanical parts, and action-level transformation that infers macroscopic cooking actions by combining the operating context of the cooking equipment. Based on the results of hierarchical semantic transformation, a motion state vector is generated.

[0101] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Determine the recipe step currently being executed by the cooking equipment and the corresponding context information for that recipe step; Based on the context information of the recipe process and multi-source heterogeneous sensor data, the type of cooking action currently being performed is determined according to a preset mapping relationship; Write the cooking action type as an attribute parameter into the motion state vector.

[0102] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Obtain the motor current from the standardized physical quantities, as well as the no-load current, rated current, rated speed, and slip ratio of the motor under no-load conditions; When the motor current is greater than the set multiple of the no-load current, the load rate is calculated based on the ratio of the difference between the motor current and the no-load current to the difference between the rated current and the no-load current. Estimate the current speed based on the load rate, rated speed, and slip rate.

[0103] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Call the pre-built parametric model library to obtain a superset skeletal tree covering the motion degrees of freedom of all models; Based on the configuration file corresponding to the model identifier of the cooking equipment, enable / disable the bone nodes in the superset skeleton tree and adjust their geometric parameters; Based on the adjusted skeletal tree, a three-dimensional visualization model corresponding to the cooking equipment is dynamically assembled and generated.

[0104] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Obtain the arrival time interval of the sensor data frames corresponding to the motion state vector; The target animation driving mode is determined based on the time interval, and the corresponding target state parameters are calculated. The target state parameters are applied to the skeletal nodes of the pre-built 3D visualization model to update the local skeletal matrix of the 3D visualization model. The updated skeletal matrix is ​​used to render and output a 3D visualization, which drives the virtual model components to move synchronously, thus achieving real-time synchronous mapping between the operating status of the cooking equipment and the virtual model.

[0105] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: If the time interval is less than or equal to the first threshold, the target state parameters of the 3D visualization model are calculated starting from the current state of each component of the pre-built 3D visualization model and ending at the target state in the current motion state vector. If the time interval is greater than the first threshold and less than the second threshold, then multiple target motion state vectors are determined, and the target state parameters of the 3D visualization model are calculated based on the multiple target motion state vectors; wherein, the multiple motion state vectors include the current motion state vector and multiple historical motion state vectors that are close in time sequence. If the time interval is greater than or equal to the second threshold, the current state parameter of the 3D visualization model is determined as the target state parameter, and a visual cue parameter is generated; wherein, the visual cue parameter is used to indicate the display of a visual cue of data interruption on the 3D visualization model.

[0106] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Receive historical sensor data from the cooking equipment and put the historical sensor data into a memory blocking queue buffer; Multiple consumer threads consume data from the blocking queue in parallel and write it to the time-series database in batches. The latest data frame in the time series database is identified as real-time multi-source heterogeneous sensor data.

[0107] Optionally, the processor 1001 may call the monitoring program of the cooking device stored in the memory 1004, and also perform the following operations: Establish a keyframe index for historical time-series data; In response to playback control commands, the nearest keyframe is located using a binary search algorithm; The precise state of the target at any given time is calculated based on the most recent keyframes and interpolation algorithms, and then used to drive a 3D visualization model for display.

[0108] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0109] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0110] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0111] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0112] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, third, etc., does not indicate any order. These words can be interpreted as names.

[0113] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0114] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring cooking equipment, characterized in that, The method includes: Acquire multi-source heterogeneous sensor data of the cooking device during operation, wherein the multi-source heterogeneous sensor data includes electrical parameters, moving part status parameters, and temperature parameters; The physical quantities in the multi-source heterogeneous sensor data are mapped to the motion parameters of the mechanical components to obtain the motion state vector; The pre-built 3D visualization model is driven by the motion state vector to achieve real-time synchronous mapping between the operating status of the cooking equipment and the virtual model. Based on the multi-source heterogeneous sensor data and historical operating data, the operating status of each component of the cooking equipment is evaluated to obtain component-level operating status evaluation results. Based on the operational status assessment results, visual status markers are rendered on the corresponding components of the 3D visualization model to display the component-level operational status.

2. The method as described in claim 1, characterized in that, The step of mapping the physical quantities in the multi-source heterogeneous sensor data to motion parameters of mechanical components to obtain a motion state vector includes: The hierarchical semantic transformation is performed on the multi-source heterogeneous sensor data. The hierarchical semantic transformation includes signal-level transformation that maps the original physical quantities in the multi-source heterogeneous sensor data to standardized physical quantities, component-level transformation that maps the standardized physical quantities to the motion state of mechanical components, and action-level transformation that infers macroscopic cooking actions by combining the operating context of the cooking equipment. Based on the result of the hierarchical semantic transformation, a motion state vector is generated.

3. The method as described in claim 2, characterized in that, The action-level transition includes: Determine the recipe step currently being executed by the cooking device and the corresponding context information for that recipe step; Based on the context information of the recipe process and the multi-source heterogeneous sensor data, the type of cooking action currently being performed is determined according to a preset mapping relationship; The cooking action type is written as an attribute parameter into the motion state vector.

4. The method as described in claim 2, characterized in that, The step of mapping standardized physical quantities to the motion state of mechanical components at the component level includes: Obtain the motor current from the standardized physical quantities, as well as the no-load current, rated current, rated speed, and slip ratio of the motor under no-load conditions; When the motor current is greater than a set multiple of the no-load current, the load rate is calculated based on the ratio of the difference between the motor current and the no-load current to the difference between the rated current and the no-load current. The current speed is estimated based on the load rate, the rated speed, and the slip ratio.

5. The method as described in claim 1, characterized in that, The method further includes: Call the pre-built parametric model library to obtain a superset skeletal tree covering the motion degrees of freedom of all models; Based on the configuration file corresponding to the model identifier of the cooking equipment, the bone nodes in the superset skeleton tree are enabled / disabled and their geometric parameters are adjusted. Based on the adjusted skeletal tree, the three-dimensional visualization model corresponding to the cooking equipment is dynamically assembled and generated.

6. The method as described in claim 1, characterized in that, The steps of driving the pre-built 3D visualization model based on the motion state vector include: Obtain the arrival time interval of the sensor data frames corresponding to the motion state vector; The target animation driving mode is determined based on the time interval, and the corresponding target state parameters are calculated. The target state parameters are applied to the skeletal nodes of the pre-built 3D visualization model to update the local skeletal matrix of the 3D visualization model. The updated skeletal matrix is ​​used to render and output a 3D visualization, which drives the virtual model components to move synchronously, thus achieving real-time synchronous mapping between the operating status of the cooking equipment and the virtual model.

7. The method as described in claim 6, characterized in that, The step of determining the target animation driving mode based on the time interval and calculating the corresponding target state parameters includes: If the time interval is less than or equal to the first threshold, the target state parameters of the three-dimensional visualization model are calculated, starting from the current state of each component of the pre-constructed three-dimensional visualization model and ending with the target state in the current motion state vector. If the time interval is greater than a first threshold and less than a second threshold, then multiple target motion state vectors are determined, and the target state parameters of the three-dimensional visualization model are calculated based on the multiple target motion state vectors; wherein, the multiple target motion state vectors include the current motion state vector and multiple historical motion state vectors that are temporally adjacent. If the time interval is greater than or equal to the second threshold, the current state parameter of the 3D visualization model is determined as the target state parameter, and a visual cue parameter is generated; wherein, the visual cue parameter is used to indicate the display of a visual cue of data interruption on the 3D visualization model.

8. The method as described in claim 1, characterized in that, The method also includes the step of acquiring real-time multi-source heterogeneous sensor data of the cooking equipment: Receive historical sensor data from the cooking device and put the historical sensor data into a memory blocking queue buffer; Multiple consumer threads consume data from the blocking queue in parallel and write it to the time-series database in batches. The latest data frame in the time series database is identified as the real-time multi-source heterogeneous sensor data.

9. The method according to any one of claims 1 to 8, characterized in that, The method further includes: Establish a keyframe index for historical time-series data; In response to playback control commands, the nearest keyframe is located using a binary search algorithm; The precise state of the target time is calculated based on the most recent keyframe and interpolation algorithm, and then the 3D visualization model is used to display it.

10. A monitoring system for cooking equipment, characterized in that, The system includes: The data acquisition module is used to acquire multi-source heterogeneous sensor data of the cooking equipment in operation, wherein the multi-source heterogeneous sensor data includes electrical parameters, moving part status parameters and temperature parameters; The semantic conversion module is used to map the physical quantities in the multi-source heterogeneous sensor data into motion parameters of mechanical components to obtain a motion state vector. The motion drive module is used to drive a pre-built three-dimensional visualization model according to the motion state vector, so as to realize the real-time synchronous mapping between the operating status of the cooking equipment and the virtual model. The operation evaluation module is used to evaluate the operation status of each component of the cooking equipment based on the multi-source heterogeneous sensor data and historical operation data, and obtain the component-level operation status evaluation results. The visualization module is used to render visualization status markers on the corresponding components of the three-dimensional visualization model based on the operational status evaluation results, so as to display the operational status at the component level.

Images