A dishwasher and a dishwasher fault detection method

Abstract

The application discloses a dishwasher and a fault detection method of the dishwasher. The method comprises the following steps: acquiring sensing signals collected by each monitoring sensor group; extracting at least one statistical feature of the sensing signals; combining the extracted statistical features into a feature vector, inputting the feature vector into a fault detection model, and acquiring a predicted fault probability of each preset fault category; determining a predicted fault category according to the predicted fault probability; and querying a water circuit fault knowledge graph according to the predicted fault category to obtain a predicted fault root cause of a water circuit. According to the embodiment of the application, the comprehensiveness and accuracy of fault detection of the dishwasher can be improved.

Description

Technical Field

[0001] This application relates to the field of dishwasher technology, and more particularly to a dishwasher and a method for detecting dishwasher malfunctions. Background Technology

[0002] The water circuit of a dishwasher, as its core operating system, performs multiple critical functions, and its operational reliability directly determines washing performance and overall machine safety. Therefore, how to achieve accurate fault detection in the water circuit has become a key research direction in the dishwasher field.

[0003] Currently, the main method for fault detection in water circuits is the single-sensor threshold comparison method, such as detecting stalled operation by drain pump current or detecting drainage anomalies by pressure switch timing. However, this fault detection method has a limited sensing dimension, is not comprehensive, and sometimes fails to accurately pinpoint the root cause of the fault, leading to difficulties in subsequent maintenance. Summary of the Invention

[0004] This application provides a dishwasher and a method for detecting dishwasher malfunctions, which can improve the comprehensiveness and accuracy of dishwasher malfunction detection.

[0005] In a first aspect, a dishwasher is provided, comprising: a water circuit including a water inlet valve, a washing inner tank, a filter, a circulating water pump, a spray arm, a drain valve, a drain pump, and a drain pipe, for forming a washing water circulation path and a drain path; Multiple monitoring sensor groups include: a water inlet monitoring sensor group for collecting the current signal and pressure signal at the water inlet valve; a leakage monitoring sensor group for collecting the impedance signal associated with the water accumulation state at the bottom of the washing tank; a filter monitoring sensor group for collecting the pressure difference signal before and after the filter and the turbidity signal of the circulating water; a spray monitoring sensor group for collecting the rotation speed signal of the spray arm and the current signal of the circulating water pump; and a drainage monitoring sensor group for collecting the current signal and pressure signal at the drainage pump. The controller, connected to the inlet valve, filter, circulating water pump, spray arm, drain valve, drain pump, and various monitoring sensor groups, is configured as follows: Acquire the sensing signals collected by each monitoring sensor group; Extract at least one statistical feature from the sensor signal, combine the extracted statistical features into a feature vector, and input the feature vector into the fault detection model to obtain the predicted fault probability for each preset fault category. Determine the predicted fault category based on the predicted fault probability; Based on the predicted fault category, the waterway fault knowledge graph is queried to obtain the predicted root cause of the waterway circuit fault.

[0006] Secondly, a fault detection method for a dishwasher is provided, comprising: acquiring sensing signals collected by each monitoring sensor group, wherein the sensing signals include: current signal and pressure signal at the water inlet valve, impedance signal associated with the water accumulation state at the bottom of the washing tub, pressure difference signal before and after the filter and turbidity signal of the circulating water, rotation speed signal of the spray arm and current signal of the circulating water pump, current signal of the drain pump and pressure signal at the drain pump. Extract at least one statistical feature from the sensor signal, combine the extracted statistical features into a feature vector, and input the feature vector into the fault detection model to obtain the predicted fault probability for each preset fault category. Determine the predicted fault category based on the predicted fault probability; Based on the predicted fault category, the waterway fault knowledge graph is queried to obtain the predicted root cause of the waterway circuit fault.

[0007] In the above embodiments, a dishwasher and a dishwasher fault detection method are provided. This method focuses on the most frequent fault scenario of the dishwasher's water circuit. Monitoring sensor groups are set up at five key stages of the water circuit: water inlet, circulation, drainage, leakage, and filtration, to collect sensor signals. Based on the collected sensor signals, a fault detection model is used to predict the probability of multiple preset fault categories, achieving simultaneous detection of multiple faults in the water circuit and improving the comprehensiveness of dishwasher fault diagnosis. Furthermore, for the predicted fault categories, the root cause of the predicted fault can be inferred in one go using a water circuit fault knowledge graph, achieving precise root cause location and improving the accuracy of dishwasher fault diagnosis. The method includes: acquiring sensor signals collected by each monitoring sensor group; extracting at least one statistical feature from the sensor signals, combining the extracted statistical features into a feature vector, and inputting the feature vector into a fault detection model to obtain the predicted fault probability of each preset fault category; determining the predicted fault category based on the predicted fault probability; and querying the water circuit fault knowledge graph based on the predicted fault category to obtain the predicted root cause of the water circuit fault. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the water circuit structure of a dishwasher provided in an embodiment of this application; Figure 2 This is a schematic diagram illustrating the setup of a water leakage monitoring sensor group according to an embodiment of this application; Figure 3 This is a structural block diagram of a dishwasher provided in an embodiment of this application; Figure 4 This is a structural block diagram of a controller provided in an embodiment of this application; Figure 5 This is a flowchart of a dishwasher fault detection method provided in an embodiment of this application; Figure 6 This is a flowchart of a method for determining and predicting the root cause of a failure, provided in an embodiment of this application. Figure 7 This is a flowchart of another dishwasher fault detection method provided in an embodiment of this application; Figure 8 This is a flowchart of another dishwasher fault detection method provided in an embodiment of this application; Figure 9 This is a flowchart illustrating the prediction of the remaining lifespan of a drainage pump, filter, and inlet valve, as provided in an embodiment of this application. Figure 10 This is a flowchart illustrating a fault detection example for a dishwasher provided in this application. Detailed Implementation

[0009] To make the objectives and implementation methods of this application clearer, the exemplary implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings of the exemplary embodiments of this application. Obviously, the exemplary embodiments described are only some embodiments of this application, and not all embodiments.

[0010] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0011] The terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar or related objects or entities, and do not necessarily imply a specific order or sequence, unless otherwise specified. It should be understood that such terms are interchangeable where appropriate.

[0012] The terms “comprising” and “having”, and any variations thereof, are intended to cover but not exclude inclusion, for example, a product or device that includes a range of components is not necessarily limited to all of the components that are clearly listed, but may include other components that are not clearly listed or that are inherent to such product or device.

[0013] In related technologies, dishwasher water circuit fault diagnosis mainly includes the following methods: First, single-sensor threshold comparison, such as detecting stalled drain pump current or abnormal drainage via pressure switch timing; second, remote cloud-based diagnosis, uploading multi-channel data to a server for trend analysis and predictive maintenance; and third, universal power-on self-test programs for home appliances, displaying fault codes and pushing notifications via an app. These solutions share common limitations: they only report a single fault code each time, diagnostic thresholds are factory-fixed and cannot adapt to component aging, they lack knowledge accumulation and self-learning capabilities, leaks only result in passive shutdown without warning, requiring users to consult the manual for repairs, and there is no mechanism for tracing the root causes and prioritizing multiple concurrent water circuits. Therefore, this application provides a dishwasher and a dishwasher fault detection method. The dishwasher will be described in detail below.

[0014] The dishwasher provided in this application can have various implementation forms, such as countertop dishwasher, built-in dishwasher, freestanding dishwasher, sink dishwasher, or drawer dishwasher, etc., and this application does not limit it.

[0015] Figure 1 This is a schematic diagram of the water circuit structure of a dishwasher provided in an embodiment of this application. Figure 2 This is a schematic diagram of the setup of a water leakage monitoring sensor group provided in an embodiment of this application. Figure 3 This is a structural block diagram of a dishwasher provided in an embodiment of this application. See also... Figure 1 , Figure 2 and Figure 3 The dishwasher includes: The water circuit includes an inlet valve 11, a washing tank 12, a filter 13, a circulating water pump 14, a spray arm 15, a drain valve 16, a drain pump 17, and a drain pipe, which are used to form a washing water circulation path and a drainage path. Multiple monitoring sensor groups include: water inlet monitoring sensor group 21, used to collect the current signal and pressure signal at the water inlet valve 11; water leakage monitoring sensor group 22, used to collect the impedance signal associated with the water accumulation state at the bottom of the washing tank 12; filter monitoring sensor group 23, used to collect the pressure difference signal before and after the filter 13 and the turbidity signal of the circulating water; spray monitoring sensor group 24, used to collect the rotation speed signal of the spray arm 15 and the current signal of the circulating water pump 14; and drainage monitoring sensor group 25, used to collect the current signal and pressure signal at the drainage pump 17. The controller is connected to the inlet valve 11, filter 13, circulating water pump 14, spray arm 15, drain valve 16, drain pump 17, and each monitoring sensor group.

[0016] Specifically, the washing water circulation path is as follows: water entering through the inlet valve 11 flows into the bottom of the washing inner tank 12, passes through the filter 13, and is pressurized by the circulating water pump 14 and sent to the spray arm 15. The spray arm 15 sprays water onto the washing inner tank 12 to rinse the dishes. The rinsed water flows back to the bottom of the washing inner tank 12, passes through the filter 13, and is pressurized again by the circulating water pump 14 and sent to the spray arm 15, thus circulating.

[0017] Specifically, the drainage path is as follows: the water in the washing tank 12 is filtered by the filter 13 and then flows into the drain pump 17 through the drain valve 16. The drain pump 17 pressurizes the water and delivers it to the drain pipe for discharge.

[0018] Specifically, the water inlet monitoring sensor group 21 includes a current sensor and a pressure sensor. The current sensor can be connected in series in the power supply line of the water inlet valve 11 to collect the current signal when the water inlet valve 11 is working; the pressure sensor can be set in the pipeline section between the water outlet of the water inlet valve 11 and the water inlet of the washing tank 12, for example, it can be set in the water outlet pipeline of the water inlet valve 11 or at the water inlet of the washing tank 12, to collect the pressure signal when the water inlet valve 11 supplies water in the open state.

[0019] For details, see Figure 2 The water leakage monitoring sensor group 22 includes a first water leakage detection electrode 221 and a second water leakage detection electrode 222. Both the first water leakage detection electrode 221 and the second water leakage detection electrode 222 are disposed at the bottom of the washing tub 12, and are spaced apart. When water accumulates at the bottom of the washing tub 12, the first water leakage detection electrode 221 and the second water leakage detection electrode 222 can form a conductive circuit through the accumulated water, outputting an impedance signal associated with the water accumulation state.

[0020] Specifically, the filter monitoring sensor group 23 includes a differential pressure sensor and a turbidity sensor. The differential pressure sensor includes a first pressure tap and a second pressure tap. The first pressure tap is located on the inlet pipe of the filter 13, and the second pressure tap is located on the outlet pipe of the filter 13. The differential pressure sensor is used to output the differential pressure signal between the inlet and outlet of the filter 13. The turbidity sensor can be located inside the washing tank 12 or at the inlet of the circulating water pump 14, and is used to output the turbidity signal of the circulating water.

[0021] For details, please refer to [link / reference]. Figure 3 The spray monitoring sensor group 24 includes a speed sensor and a current sensor. The speed sensor can be set at the rotating shaft of the spray arm 15 or at the end of the spray arm 15 to collect the speed signal of the spray arm 15; the current sensor can be connected in series in the power supply line of the circulating water pump 14 to collect the current signal when the circulating water pump 14 is working.

[0022] Specifically, the drainage monitoring sensor group 25 includes a current sensor and a pressure sensor. The current sensor can be connected in series with the power supply line of the drainage pump 17 to collect the current signal when the drainage pump 17 is working; the pressure sensor can be set on the inlet pipe of the drainage pump 17, the outlet pipe of the drainage pump 17, or the pump body of the drainage pump 17 to collect the pressure signal at the drainage pump 17.

[0023] Figure 4 This is a structural block diagram of a controller provided in an embodiment of this application. See also... Figure 4The controller 30 includes a processor 31 and a memory 32 storing computer program instructions. The processor 31 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of this disclosure. The memory 32 may include a mass storage device for information or instructions. For example, and not limitingly, the memory 32 may include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 32 may include removable or non-removable (or fixed) media. Where appropriate, the memory 32 may be internal or external to the integrated gateway device. In a particular embodiment, the memory 32 is a non-volatile solid-state memory. In a particular embodiment, the memory 32 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (Electrically Programmable ROM, EPROM), an electrically erasable PROM (EEPROM), an electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these. The processor 31 executes the steps of the dishwasher fault detection method provided in this disclosure by reading and executing computer program instructions stored in the memory 32. In one example, the controller 30 may further include a transceiver 33 and a bus 34. As shown in the figures, the processor 31, memory 32, and communication interface are connected via the bus 34 and communicate with each other. The communication interface can be used to communicate with other devices or communication networks (such as Ethernet, radio access network (RAN)), wireless local area networks (WLAN), etc. The communication interface may be a module, circuit, transceiver, or any device capable of communication. Bus 34 includes peripheral component interconnect (PCI) lines or extended industry standard architecture (EISA) buses, etc.Bus 34 can be divided into address bus, data bus, control bus, etc.

[0024] In some embodiments, the controller is configured to acquire sensing signals collected by each monitoring sensor group; Extract at least one statistical feature from the sensor signal, combine the extracted statistical features into a feature vector, and input the feature vector into the fault detection model to obtain the predicted fault probability for each preset fault category. Determine the predicted fault category based on the predicted fault probability; Based on the predicted fault category, the waterway fault knowledge graph is queried to obtain the predicted root cause of the waterway circuit fault.

[0025] In some embodiments, when determining the predicted fault category based on the predicted fault probability, the controller is specifically configured to: For each preset fault category, when the predicted fault probability of the preset fault category is greater than the preset probability threshold corresponding to the preset fault category N times consecutively, the preset fault category is confirmed as a predicted fault category.

[0026] In some embodiments, when querying a waterway fault knowledge graph based on the predicted fault category to obtain the predicted root cause of a waterway loop fault, the controller is specifically configured as follows: For each predicted fault category, the waterway fault knowledge graph is queried to obtain the candidate root causes of the predicted fault category; Calculate the intersection of candidate root causes for each predicted fault category; If the intersection is not an empty set, then the candidate root causes in the intersection will be used as the predicted root causes of the failure. If the intersection is an empty set, then all candidate root causes for each predicted fault category will be used as predicted fault root causes.

[0027] In some embodiments, after querying the waterway fault knowledge graph based on the predicted fault category to obtain the predicted root cause of the waterway loop, the controller is further configured to: Based on the predicted root causes of the faults, the waterway fault knowledge graph is queried to obtain the predicted fault repair strategy for the waterway loop. It outputs to users the predicted fault category, the predicted root cause of the fault, and the predicted fault repair strategy.

[0028] In some embodiments, after outputting the predicted fault category, the predicted root cause of the fault, and the predicted fault repair strategy to the user, the controller is further configured to: If the predicted fault repair strategy is an automatic execution strategy, then the corresponding target device in the dishwasher will be controlled to perform fault repair operations according to the predicted fault repair strategy. If the predicted fault repair strategy is the user's execution strategy, then the fault repair guidance will be output to the user in a multi-turn dialogue manner according to the predicted fault repair strategy. After the predicted fault repair strategy is completed, the sensor signals collected by each monitoring sensor group are reacquired, and the predicted fault categories are determined based on the reacquired sensor signals. If there are residual fault categories that have not been repaired among the predicted fault categories, the corresponding fault repair strategy will be re-executed for the residual fault categories until all predicted fault categories have been repaired or a fault repair completion notification is received from the user.

[0029] In some embodiments, after all predicted fault categories have been repaired or a fault repair completion notification input by the user has been received, the controller is further configured to: Obtain the actual fault category input by the user; For each actual fault category, if the actual fault category is one of the preset fault categories, the predicted fault probability corresponding to the actual fault category is determined as the actual fault probability; if the actual fault category is not one of the preset fault categories, the actual fault probability of the actual fault category is determined as the pre-configured probability. The actual fault category, actual fault probability, and feature vector are stored in the historical fault database as historical fault samples. Based on the matching between the actual fault categories and each predicted fault category, the deviation fault category is determined; For each type of deviation fault, the difference between the actual fault probability and the predicted fault probability of the deviation fault category is calculated to obtain the deviation fault probability. If the probability of a deviation fault in at least one deviation fault category is greater than a preset deviation threshold, the fault detection model is retrained using historical fault samples.

[0030] In some embodiments, the controller is further configured to: extract the actual fault probability of the preset fault category from the historical fault database for each preset fault category, and adjust the preset probability threshold corresponding to the preset fault category according to the extracted actual fault probability.

[0031] In some embodiments, after all predicted fault categories have been repaired or a fault repair completion notification input by the user has been received, the controller is further configured to: Obtain the actual fault category, actual fault root cause, and actual fault repair strategy input by the user; Update the waterway fault knowledge graph based on the actual fault type, the actual fault root cause, and the actual fault repair strategy.

[0032] In some embodiments, the controller is further configured to: determine the water inlet time required for the water inlet valve to open and close based on the pressure signal at the water inlet valve, and calculate the remaining lifespan of the water inlet valve based on the water inlet time; Based on the pressure signal at the drain pump, determine the drainage time required from start-up to end of drainage, and calculate the remaining life of the drain pump based on the drainage time. Calculate the remaining lifespan of the filter based on the pressure difference signals before and after the filter; Based on the remaining lifespan of the inlet valve, the drain pump, and the filter, the corresponding warning levels for the inlet valve, drain pump, and filter are determined, and corresponding warning messages are output according to the warning levels.

[0033] Figure 5 This is a flowchart of a dishwasher fault detection method provided in an embodiment of this application. See also... Figure 5 The method includes the following steps: S110. Acquire the sensing signals collected by each monitoring sensor group.

[0034] Specifically, the sensor signals include the current signal of the inlet valve, the pressure signal at the inlet valve, the impedance signal related to the water accumulation state at the bottom of the washing tank, the pressure difference signal before and after the filter, the turbidity signal of the circulating water, the speed signal of the spray arm, the current signal of the circulating water pump, the current signal of the drain pump, and the pressure signal at the drain pump.

[0035] S120. Extract at least one statistical feature from the sensor signal, combine the extracted statistical features into a feature vector, and input the feature vector into the fault detection model to obtain the predicted fault probability for each preset fault category.

[0036] Specifically, the preset fault categories may include: abnormal water inlet, water leakage in the washing tank, filter blockage, spray arm blockage, and drainage failure. Further optionally, abnormal water inlet may include insufficient water flow from the inlet valve and / or excessively long water inlet time; filter blockage may include excessive pressure difference across the filter and / or reduced filtration capacity; spray arm blockage may include blockage at the spray arm outlet and / or abnormal rotation speed; and drainage failure may include poor drainage from the drain pump and / or excessively long drainage time.

[0037] Specifically, for the sensing signals collected by each monitoring sensor group, statistical features (such as mean, variance, peak value, etc.) are extracted independently. Then, the statistical features corresponding to each monitoring sensor group are concatenated into a comprehensive vector to obtain a feature vector. The feature vector is input into the fault detection model for concurrent inference, and the predicted fault probability for each preset fault category is output.

[0038] Specifically, there are various training methods for the fault detection model, and this application does not limit them. For example, the sensor signals collected by each monitoring sensor group of the dishwasher under the following conditions are obtained: normal state, abnormal water inlet, water leakage in the washing tank, filter blockage, spray arm blockage, and drainage failure. The sensor signals are processed to obtain feature vectors, and the corresponding fault category labels are labeled on the sensor signals. The initial fault detection model is iteratively trained using the feature vectors as input and the fault category labels as supervision until the model converges.

[0039] For example, after the feature vector is input into the fault detection model, the predicted fault probability is 0.05 for water inlet abnormality, 0.02 for water leakage in the washing tank, 0.82 for filter blockage, 0.68 for spray arm blockage, and 0.75 for drainage failure.

[0040] S130. Determine the predicted fault category based on the predicted fault probability.

[0041] In some embodiments, S130 includes: for each preset fault category, when the predicted fault probability of the preset fault category is greater than the preset probability threshold corresponding to the preset fault category, confirming the preset fault category as a predicted fault category. Further optionally, S130 includes: for each preset fault category, when the predicted fault probability of the preset fault category is greater than the preset probability threshold corresponding to the preset fault category N times consecutively, confirming the preset fault category as a predicted fault category.

[0042] Specifically, N is a positive integer. Those skilled in the art can set the specific value of N according to the actual situation, and this application does not limit it in this regard. For example, N can take any value in the range of 2 to 5, such as N being 3, but it is not limited to this.

[0043] For example, if the predicted failure probability of a drainage failure is 0.75 in the first sampling (exceeding the threshold of 0.6), and is 0.78 in the second sampling and 0.76 in the third sampling, exceeding the corresponding preset probability threshold three times in a row, then the drainage failure is confirmed as a predicted failure category; if it drops to 0.55 in the second sampling, then the drainage failure is not a predicted failure category, and monitoring can continue.

[0044] Understandably, setting a preset probability threshold independently for each preset fault category allows for configuring detection sensitivity based on the fluctuation characteristics of different preset fault categories, avoiding missed or false alarms caused by a uniform probability threshold. Furthermore, the system requires N consecutive exceedances of the corresponding preset probability threshold before confirming a predicted fault category, effectively filtering out signal spikes caused by transient fluctuations in the water circuit, reducing the false alarm rate, and improving the reliability of fault diagnosis.

[0045] In some embodiments, S130 includes: for each preset fault category, when the predicted fault probability of the preset fault category is greater than a fixed probability threshold, confirming the preset fault category as a predicted fault category. That is, all preset fault categories share the same fixed probability threshold.

[0046] S140. Based on the predicted fault category, query the waterway fault knowledge graph to obtain the predicted root cause of the waterway circuit fault.

[0047] Specifically, the waterway fault knowledge graph is a structured knowledge base, including multiple types of nodes and causal relationship edges connecting these nodes. The multiple types of nodes include: symptom nodes, cause nodes, treatment plan nodes, device nodes, and waterway location nodes.

[0048] Symptom nodes are used to characterize the fault categories of the water circuit, including: abnormal water inlet, water leakage in the washing tank, filter blockage, spray arm blockage, and drainage failure.

[0049] Root cause nodes are used to characterize the root causes that lead to the malfunctions corresponding to symptom nodes. For example, root causes associated with the water inlet abnormality node include: damaged water inlet valve, insufficient water pressure; root causes associated with the washing tank leakage node include: aging door seal, loose drain pipe; root causes associated with the filter blockage node include: long-term uncleaning, grease buildup, food residue accumulation; root causes associated with the spray arm blockage node include: blocked spray arm outlet, insufficient water pressure in the circulating water pump, blocked filter; root causes associated with the drainage failure node include: aging drain pump, blocked drain pipe, blocked filter.

[0050] Treatment plan nodes are used to characterize the repair strategies for eliminating the faults (or the root causes) corresponding to the symptom nodes. For example, repair strategies associated with the water inlet abnormality node include: replacing the water inlet valve and checking the water inlet pressure; repair strategies associated with the washing tank leakage node include: replacing the door seal and repairing the drain pipe; repair strategies associated with the filter blockage node include: cleaning the filter, cleaning with a special cleaning agent, and replacing the filter; repair strategies associated with the spray arm blockage node include: cleaning the spray arm outlet, repairing the circulating water pump, and cleaning the filter; and repair strategies associated with the drainage failure node include: replacing the drain pump, unclogging the drain pipe, and cleaning the filter.

[0051] Device nodes are used to characterize devices associated with the faults corresponding to symptom nodes.

[0052] Waterway location nodes are used to characterize the location of the fault corresponding to the symptom node in the waterway loop.

[0053] The causal relationship edges are directed edges, including directed edges from symptom nodes to cause nodes, treatment plan nodes, device nodes, and waterway location nodes, used to characterize the causal relationship between symptom nodes and cause nodes, treatment plan nodes, device nodes, and waterway location nodes; they may also include causal relationships from cause nodes to treatment plan nodes.

[0054] Figure 6 This is a flowchart illustrating how to determine a predicted root cause of a failure, as provided in an embodiment of this application. See also... Figure 6 Optionally, S140 includes: S141, for each predicted fault category, querying the waterway fault knowledge graph to obtain candidate root causes of the predicted fault category.

[0055] Specifically, for each predicted fault category, the waterway fault knowledge graph is queried, and the root cause in the root cause node pointed to by the predicted fault category is taken as the candidate root cause of the predicted fault category.

[0056] S142. Calculate the intersection of candidate root causes for each predicted fault category.

[0057] Specifically, the candidate root causes for each predicted fault category constitute a set of candidate root causes; the intersection of the candidate root cause sets corresponding to each predicted fault category is calculated.

[0058] S143. If the intersection is not an empty set, then the candidate root causes in the intersection will be used as the predicted root causes of the failure.

[0059] S144. If the intersection is an empty set, then all candidate root causes of each predicted fault category will be used as predicted fault root causes.

[0060] Specifically, if the intersection is not an empty set, it indicates that there is a common root cause among the candidate root cause sets corresponding to each predicted fault category. That is, the faults of the multiple predicted fault categories that are currently occurring are not independent of each other, but are caused by the same common root cause. At this time, it can be determined that the dishwasher has an associated fault, and the root cause in the intersection is taken as the predicted fault root cause.

[0061] Specifically, if the intersection is an empty set, it indicates that there is no common root cause among the candidate root cause sets corresponding to each predicted fault category. That is, the multiple predicted fault categories that are currently occurring are caused by their own independent root causes, and there is no causal relationship between them. In this case, it can be determined that the dishwasher has multiple independent faults, and the candidate root causes of each predicted fault category are all taken as predicted fault root causes.

[0062] For example, if the predicted fault categories include drainage fault, spray arm blockage, and filter blockage, querying the water system fault knowledge graph yields the following candidate root causes: The candidate root cause set for drainage fault is: aging drain pump, blocked drain pipe, blocked filter; the candidate root cause set for spray arm blockage is: blocked spray arm outlet, insufficient water pressure in the circulating water pump, blocked filter; the candidate root cause set for filter blockage is: long-term uncleaning, grease buildup, food residue accumulation. The intersection of these three candidate root cause sets is filter blockage, identifying the associated fault in the dishwasher, and using filter blockage as the predicted root cause.

[0063] For example, if the predicted fault categories include water inlet abnormality and washing tank leakage, querying the water circuit fault knowledge graph yields the following candidate root causes: The candidate root cause set for water inlet abnormality includes: damaged water inlet valve and insufficient water pressure; the candidate root cause set for washing tank leakage includes: aging door seal and loose drain pipe. The intersection of the two candidate root cause sets is empty, indicating that the dishwasher has multiple independent faults, and the damaged water inlet valve, insufficient water pressure, aging door seal, and loose drain pipe are all considered as predicted fault root causes.

[0064] Understandably, by calculating the intersection of the candidate root cause sets for each predicted fault category, it is possible to determine whether multiple faults share a common root cause. If the intersection is not empty, the faults are identified as related, allowing subsequent repairs based on the common root cause. Users only need to perform the repair operation corresponding to the common root cause to eliminate all faults, avoiding the time wasted on repeated repairs. If the intersection is empty, the faults are identified as independent multiple faults, allowing subsequent repairs based on the candidate root causes of each predicted fault category one by one, preventing some faults from being left unrepaired.

[0065] This application focuses on the most frequent fault scenario in the dishwasher's water circuit. Monitoring sensor groups are installed at five key stages of the water circuit: water inlet, circulation, drainage, leakage, and filtration. Based on these sensor signals, a fault detection model is used to predict the probability of various preset fault categories, achieving simultaneous detection of multiple faults in the water circuit and improving the comprehensiveness of dishwasher fault diagnosis. Furthermore, for the predicted fault categories, a water circuit fault knowledge graph can be used to deduce the predicted root cause in one go, achieving precise root cause location and improving the accuracy of dishwasher fault diagnosis.

[0066] Figure 7 This is a flowchart of another dishwasher fault detection method provided in an embodiment of this application. See also... Figure 7 Following S140, the following steps are also included: S150, based on the predicted root cause of the fault, query the water circuit fault knowledge graph to obtain the predicted fault repair strategy for the water circuit loop, and output the predicted fault category, predicted fault root cause, and predicted fault repair strategy to the user.

[0067] Specifically, for each predicted root cause of a fault, the waterway fault knowledge graph is queried, and the repair strategy in the treatment plan node pointed to by the predicted root cause is used as the predicted fault repair strategy for the predicted root cause.

[0068] Specifically, when the intersection of the candidate root cause sets corresponding to each predicted fault category is not empty, the predicted fault category, common root cause, and the corresponding repair strategy (i.e., the predicted fault repair strategy) are output through voice, text, or other means.

[0069] Specifically, when the intersection of the candidate root cause sets corresponding to each predicted fault category is an empty set, the predicted fault category, the predicted root cause of each predicted fault category, and the corresponding remediation strategy (i.e., the predicted fault remediation strategy) are output via voice, text, or other means. Alternatively, the predicted fault remediation strategies can be output sequentially according to the security level corresponding to each predicted fault category, from highest to lowest.

[0070] Optionally, based on the predicted fault category, the water circuit fault knowledge graph can be queried to obtain the predicted faulty device in the water circuit, and then a simplified diagram of the dishwasher water circuit can be displayed on the screen with the faulty device's location flashing.

[0071] For example, as mentioned above, if the common root cause is filter blockage, and the water circuit fault knowledge graph is consulted to obtain the predicted fault repair strategy for the water circuit loop as cleaning the filter, then a simplified diagram of the water circuit loop can be displayed on the screen, with the filter location flashing red and the text displaying "Filter blockage"; at the same time, a voice announcement can be made: "Filter blockage detected, which affects both drainage and spraying effects. Simply cleaning the filter will solve the problem."

[0072] Understandably, by providing users with predicted fault categories, predicted root causes, and predicted fault repair strategies, users can clearly understand the current fault categories in the water circuit, the root causes of the faults, and the corresponding solutions. This reduces the difficulty and time cost for users to troubleshoot faults themselves, and improves the accuracy and efficiency of fault repair.

[0073] Figure 8 This is a flowchart of another dishwasher fault detection method provided in an embodiment of this application. See also... Figure 8 Following S150, the following steps are also included: S160, if the predicted fault repair strategy is an automatic execution strategy, then the corresponding target device in the dishwasher is controlled to perform fault repair operations according to the predicted fault repair strategy.

[0074] Specifically, the automatic execution strategy refers to a predictive fault repair strategy where all repair operations do not require manual user intervention. Therefore, the controller can determine at least one target device to be controlled based on the predictive fault repair strategy, generate control commands to drive the target device to perform repair operations, and send these control commands to the target device to enable it to execute the repair operations.

[0075] For example, if the predicted fault repair strategy is to clean the filter, the controller can automatically execute a backwashing procedure (such as backwashing the filter with high-pressure water for 30 seconds).

[0076] S170. If the predicted fault repair strategy is a user-executed strategy, then output fault repair guidance to the user in a multi-turn dialogue manner according to the predicted fault repair strategy.

[0077] Specifically, troubleshooting instructions are information used to guide users to perform repair operations. This may include, for example, repair procedures that require manual execution by the user, information on components that the user needs to check or replace, and / or operation prompts that require user confirmation, but is not limited to these.

[0078] Specifically, the user execution strategy refers to the fact that at least one repair operation in the predicted fault repair strategy requires manual operation by the user. Therefore, the controller can gradually provide fault repair guidance to the user through voice or text in a multi-turn dialogue, thereby guiding the user to manually repair the fault. Of course, the depth of guidance can be adapted according to the user profile when providing fault repair guidance to the user.

[0079] For example, if the predicted fault repair strategy is to clean the spray arm outlet, the controller can provide voice guidance: Please remove the spray arm and check if the outlet is blocked. Do you need more detailed fault repair instructions? If the user answers: Yes, then continue the guidance: Pull the spray arm upwards to remove it, rinse the outlet with clean water, and then reinstall it.

[0080] S180. After the predicted fault repair strategy is completed, the sensor signals collected by each monitoring sensor group are reacquired, and the predicted fault categories are determined based on the reacquired sensor signals.

[0081] Specifically, after the predicted fault repair strategy is executed, the sensor signals collected by each monitoring sensor group are reacquired, and the predicted fault probability of each preset fault category is reacquired based on the reacquired sensor signals. Based on the reacquired predicted fault probability, it is determined whether there are any residual fault categories that have not been repaired.

[0082] S190. If there are residual fault categories that have not been repaired in the predicted fault categories, the corresponding fault repair strategy shall be re-executed for the residual fault categories until all predicted fault categories are repaired or a fault repair completion notification input by the user is obtained.

[0083] Specifically, for residual fault categories, if the predicted fault repair strategy corresponding to the residual fault category is an automatic execution strategy, then the corresponding target device in the dishwasher will be controlled to perform fault repair operations according to the predicted fault repair strategy; if the predicted fault repair strategy corresponding to the residual fault category is a user execution strategy, then more detailed fault repair guidance can be output to the user to improve the repair success rate.

[0084] Specifically, if there are no unrepaired residual fault categories in the predicted fault categories, a prompt message indicating that all faults have been eliminated will be output via voice or text.

[0085] It can be understood that by performing repair verification, the integrity and reliability of fault repair are ensured, and the recurrence of faults due to omissions or incomplete repairs is avoided.

[0086] Optionally, after all predicted fault categories have been repaired or a fault repair completion notification input by the user has been received, the method further includes: obtaining the actual fault category input by the user; For each actual fault category, if the actual fault category is one of the preset fault categories, the predicted fault probability corresponding to the actual fault category is determined as the actual fault probability; if the actual fault category is not one of the preset fault categories, the actual fault probability of the actual fault category is determined as the pre-configured probability. The actual fault category, actual fault probability, and feature vector are stored in the historical fault database as historical fault samples. Based on the matching between the actual fault categories and each predicted fault category, the deviation fault category is determined; For each type of deviation fault, the difference between the actual fault probability and the predicted fault probability of the deviation fault category is calculated to obtain the deviation fault probability. If the probability of a deviation fault in at least one deviation fault category is greater than a preset deviation threshold, the fault detection model is retrained using historical fault samples.

[0087] Specifically, the actual fault category is used to characterize the fault category that actually occurred after user confirmation.

[0088] Specifically, when the actual fault category belongs to one of the predicted fault categories, the predicted fault probability output by the fault detection model for that predicted fault category is determined as the actual fault probability for that actual fault category. When the actual fault category does not belong to any of the predicted fault categories, i.e., when the actual fault category is a fault category missed by the fault detection model, the actual fault probability of that actual fault category is determined as a pre-configured probability. This pre-configured probability is a fault probability pre-configured for "actual fault categories that do not belong to predicted fault categories," indicating the presence of a fault of that actual fault category in the water circuit. Furthermore, the actual fault category, actual fault probability, and feature vector are stored in a historical fault database as historical fault samples for subsequent fine-tuning of the fault detection model.

[0089] Specifically, when the actual fault category does not belong to any of the predicted fault categories, the actual fault category is determined as a deviation fault category; when the predicted fault category does not belong to any of the actual fault categories, the predicted fault category is determined as a deviation fault category.

[0090] Specifically, the system reads each historical fault sample stored in the historical fault database; uses the feature vector of each historical fault sample as input and the corresponding actual fault label in each historical fault sample as a supervision signal to fine-tune the fault detection model and update the model parameters in the fault detection model.

[0091] It is understandable that by updating the fault detection model, the model can continuously learn and optimize using the ever-accumulating historical fault samples, thereby improving the fault identification capability.

[0092] Optionally, the method further includes: for each preset fault category, extracting the actual fault probability of the preset fault category from the historical fault database, and adjusting the preset probability threshold corresponding to the preset fault category according to the extracted actual fault probability.

[0093] Specifically, the preset probability thresholds corresponding to preset fault categories can be updated by drift. For example, for each preset fault category, the actual fault probability corresponding to that preset fault category is extracted from the historical fault database, the mean of the actual fault probability is calculated, the difference between the current preset probability threshold corresponding to that preset fault category and the mean is calculated, the difference is multiplied by a preset drift rate coefficient to obtain the drift increment, and then the current preset probability threshold is added to the drift increment to obtain the updated preset probability threshold, but it is not limited to this.

[0094] Of course, the update of the fault detection model and the preset probability threshold can be controlled by the learning rate, which gradually decreases as the dishwasher's service life increases.

[0095] Understandably, by updating the preset probability threshold, the preset probability threshold can be adapted in a timely manner to the drift caused by factors such as component aging and environmental changes in the dishwasher, thereby improving the fault identification capability.

[0096] Optionally, after all predicted fault categories have been repaired or a fault repair completion notification input by the user has been received, the method further includes: obtaining the actual fault category, the actual fault root cause, and the actual fault repair strategy input by the user. Update the waterway fault knowledge graph based on the actual fault type, the actual fault root cause, and the actual fault repair strategy.

[0097] Specifically, the actual root cause of a fault is used to characterize the fundamental cause, as confirmed by the user, that leads to an actual fault in the water circuit. The actual fault repair strategy is used to characterize the repair strategy actually implemented by the user to eliminate the fault of the actual fault category.

[0098] Specifically, the system retrieves the corresponding symptom node based on the actual fault category, updates the cause node pointed to by the symptom node based on the actual root cause of the fault, and updates the treatment plan node pointed to by the cause node based on the actual fault repair strategy. Similarly, it can also obtain the actual faulty device and the actual location of the fault, and update the water circuit fault knowledge graph based on these information. This allows for continuous optimization of the water circuit fault knowledge graph, providing more accurate repair strategies and thus improving repair efficiency.

[0099] Of course, the actual root cause of the fault, the actual fault repair strategy, the repair result, the repair time, and user satisfaction can also be stored in the historical fault database. For example, the fault repair record for this time is as follows: the feature vector includes the drain pump current signal of 1.8A, the pressure difference before and after the filter of 18kPa, and the turbidity signal of the circulating water of 800NTU; the predicted fault category is filter blockage, and the predicted fault probability is 0.8; the actual fault category is filter blockage, and the actual fault probability is 0.8; the predicted root cause is filter blockage, the actual root cause is filter blockage, the predicted fault repair strategy is cleaning the filter, the actual fault repair strategy is cleaning the filter, the repair result is successful, the repair time is 8 minutes, and the user satisfaction is satisfied.

[0100] Figure 9 This is a flowchart illustrating the prediction of the remaining lifespan of a drain pump, filter, and inlet valve, as provided in an embodiment of this application. See also... Figure 9 Optionally, the method further includes: S210, determining the water inlet time required from opening to closing of the water inlet valve based on the pressure signal at the water inlet valve, and calculating the remaining life of the water inlet valve based on the water inlet time.

[0101] Specifically, the time from the start of water intake to the point when the pressure signal at the inlet valve rises to a preset pressure threshold is the water intake time. An inlet valve degradation model can be pre-established. In this model, a shorter water intake time corresponds to a longer remaining lifespan, and vice versa. The water intake time is plotted on the x-axis, and the remaining lifespan of the inlet valve is plotted on the y-axis. Based on the inlet valve degradation model, the remaining lifespan of the inlet valve can be calculated.

[0102] S220. Based on the pressure signal at the drain pump, determine the drainage time required for the drain pump to start and finish, and calculate the remaining life of the drain pump based on the drainage time.

[0103] Specifically, the drainage time is the difference between the start of drainage and the point at which the pressure signal at the drainage pump drops to a preset pressure threshold. A drainage pump degradation model can be pre-established, in which shorter drainage time corresponds to a longer remaining lifespan, and vice versa. Drainage time is plotted on the x-axis, and the remaining lifespan of the drainage pump is plotted on the y-axis. Based on the drainage pump degradation model, the remaining lifespan of the drainage pump can be calculated.

[0104] S230. Calculate the remaining life of the filter based on the pressure difference signal before and after the filter.

[0105] Specifically, a filter degradation model can be pre-established. In this model, a smaller pressure difference across the filter results in a longer remaining lifespan, while a larger pressure difference results in a shorter remaining lifespan. The pressure difference is plotted on the x-axis, and the remaining lifespan of the filter is plotted on the y-axis. Based on the filter degradation model, the remaining lifespan of the filter can be calculated.

[0106] S240. Based on the remaining lifespan of the inlet valve, the remaining lifespan of the drain pump, and the remaining lifespan of the filter, determine the corresponding warning level for the inlet valve, the drain pump, and the filter, and output the corresponding warning message according to the warning level.

[0107] Specifically, multiple warning levels can be set, with different warning messages sent to the user at different levels. For example, if the remaining lifespan is less than the first preset number of days threshold (e.g., 7 days), it is classified as the first warning level (red warning); if the remaining lifespan is greater than or equal to the first preset number of days threshold and less than the second preset number of days threshold (e.g., 30 days), it is classified as the second warning level (red warning); and if the remaining lifespan is greater than or equal to the second preset number of days threshold and less than the third preset number of days threshold (e.g., 90 days), it is classified as the third warning level (green warning).

[0108] Understandably, by predicting the remaining lifespan of the inlet valve, drain pump, and filter and providing tiered warnings, users can learn about the health status of the components before they completely fail, thus allowing them to plan maintenance in advance and avoid dishwasher downtime and emergency repair costs due to sudden malfunctions.

[0109] To illustrate the dishwasher fault detection method provided in this application in detail, a specific example is given below. For example, Figure 10 This is a flowchart illustrating a fault detection example for a dishwasher provided in this application. See also... Figure 10This application incorporates multiple monitoring sensor groups in the water circuit, including: an inlet water monitoring sensor group, a leakage monitoring sensor group, a filter monitoring sensor group, a spray monitoring sensor group, and a drainage monitoring sensor group. The inlet water monitoring sensor group monitors for abnormalities during the water inlet stage and may include a flow sensor to collect the flow signal downstream of the inlet valve and a level sensor to collect the liquid level signal in the washing tank. It may also include a current sensor to collect the current signal from the inlet valve and a pressure sensor to collect the pressure signal at the inlet valve. The filter monitoring sensor group includes a pressure sensor to collect the pressure difference signal before and after the filter, a turbidity sensor to collect the turbidity signal of the circulating water, and a current sensor to collect the current signal of the circulating water pump. This is used to identify circulating washing faults through pressure and current linkage monitoring and to provide feedback on filter blockage status through turbidity feedback. The drainage monitoring sensor group includes a level sensor to collect the liquid level signal in the washing tank, used to determine whether drainage is normal based on the liquid level and the rate of liquid level drop. Blockage detection can also include current sensors that collect the drain pump's current signal and pressure sensors that collect the pressure signal at the drain pump. This allows for the determination of whether drainage is blocked or abnormal by coordinating the rate of liquid level drop with pressure and current. A leak detection sensor group, including a first and second leak detection electrode, collects impedance signals associated with the water accumulation at the bottom of the washing tank, serving as redundant overflow prevention and triggering protection when the main liquid level fails. A spray monitoring sensor group, including a speed sensor that collects the spray arm's rotation speed signal and the circulating water pump's current signal, monitors whether the spray arm is stuck or idling. Thus, by combining a fault detection model with a water circuit fault knowledge graph, real-time fault perception and accurate detection can be achieved across multiple operating stages, including water intake, washing, and drainage. When a fault is detected, a predicted fault repair strategy is determined based on the water circuit fault knowledge graph. This strategy is then executed automatically, or the user is guided to manually handle the fault through human-machine interaction. When an abnormal water intake is detected, a pulse on / off reset operation is performed. If self-healing fails after reset, the user is notified to check the water intake path or contact after-sales service, and the corresponding fault code is output. When a cycle fault is detected, operation is paused and restarted, while simultaneously activating the speed compensation mechanism. If the fault is not resolved, operation is downshifted (or the process switches to drainage), and the corresponding fault code is output. When drainage blockage is detected, a forward and reverse clearing strategy is initiated (5 seconds forward, 2 seconds reverse). If blockage is still detected, drainage time is extended (by 50%). If the process still fails after extension, the fault is reported and after-sales service is notified, and the corresponding fault code is output. When overflow risk is detected, overflow forced drainage linkage protection is immediately activated (water cut-off, power cut-off, and forced drainage are executed). If the liquid level does not effectively drop, forced drainage continues until the machine is finally locked and shut down, and the corresponding fault code is output.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0111] For ease of explanation, the above description has been provided in conjunction with specific embodiments. However, the above exemplary discussion is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. Various modifications and variations can be obtained based on the above teachings. The selection and description of the above embodiments are for the purpose of better explaining the principles and practical applications, thereby enabling those skilled in the art to better utilize the described embodiments and various different variations of embodiments suitable for specific use considerations.

Claims

1. A dishwasher, characterized in that, include: The water circuit includes an inlet valve, a washing tank, a filter, a circulating water pump, a spray arm, a drain valve, a drain pump, and a drain pipe, which are used to form the washing water circulation path and the drainage path. Multiple monitoring sensor groups include: a water inlet monitoring sensor group for collecting the current signal and pressure signal at the water inlet valve; a leakage monitoring sensor group for collecting the impedance signal associated with the water accumulation state at the bottom of the washing tank; a filter monitoring sensor group for collecting the pressure difference signal before and after the filter and the turbidity signal of the circulating water; a spray monitoring sensor group for collecting the rotation speed signal of the spray arm and the current signal of the circulating water pump; and a drainage monitoring sensor group for collecting the current signal and pressure signal at the drainage pump. The controller, connected to the inlet valve, the filter, the circulating water pump, the spray arm, the drain valve, the drain pump, and each of the monitoring sensor groups, is configured as follows: Acquire the sensing signals collected by each of the monitoring sensor groups; At least one statistical feature of the sensing signal is extracted, the extracted statistical features are combined into a feature vector, and the feature vector is input into the fault detection model to obtain the predicted fault probability of each preset fault category. Based on the predicted failure probability, the predicted failure category is determined; Based on the predicted fault category, the waterway fault knowledge graph is queried to obtain the predicted root cause of the fault in the waterway loop.

2. The dishwasher according to claim 1, characterized in that, When determining the predicted fault category based on the predicted fault probability, the controller is specifically configured as follows: For each of the preset fault categories, when the predicted fault probability of the preset fault category is greater than the preset probability threshold corresponding to the preset fault category for N consecutive times, the preset fault category is confirmed as the predicted fault category.

3. The dishwasher according to claim 1, characterized in that, When querying the waterway fault knowledge graph based on the predicted fault category to obtain the predicted root cause of the waterway loop, the controller is specifically configured as follows: For each of the predicted fault categories, the waterway fault knowledge graph is queried to obtain the candidate root causes of the predicted fault category; Calculate the intersection of the candidate root causes for each of the predicted fault categories; If the intersection is not an empty set, then the candidate root causes in the intersection are taken as the predicted root causes of the failure. If the intersection is an empty set, then all candidate root causes of each predicted fault category are taken as the predicted fault root causes.

4. The dishwasher according to claim 3, characterized in that, After querying the waterway fault knowledge graph based on the predicted fault category to obtain the predicted root cause of the waterway loop, the controller is further configured to: Based on the predicted root cause of the fault, the waterway fault knowledge graph is queried to obtain the predicted fault repair strategy for the waterway loop. The system outputs the predicted fault category, the predicted fault root cause, and the predicted fault repair strategy to the user.

5. The dishwasher according to claim 4, characterized in that, After outputting the predicted fault category, the predicted fault root cause, and the predicted fault repair strategy to the user, the controller is further configured to: If the predicted fault repair strategy is an automatic execution strategy, then the corresponding target device in the dishwasher is controlled to perform fault repair operations according to the predicted fault repair strategy; If the predicted fault repair strategy is a user-executed strategy, then fault repair guidance will be output to the user in a multi-turn dialogue manner according to the predicted fault repair strategy. After the predicted fault repair strategy is completed, the sensing signals collected by each of the monitoring sensor groups are reacquired, and it is determined whether the predicted fault categories have been repaired based on the reacquired sensing signals. If there are any residual fault categories among the predicted fault categories that have not been repaired, the corresponding fault repair strategy will be re-executed for the residual fault categories until all the predicted fault categories have been repaired or a fault repair completion notification input by the user is received.

6. The dishwasher according to claim 5, characterized in that, After all predicted fault categories have been repaired or a fault repair completion notification has been received from the user, the controller is further configured to: Obtain the actual fault category input by the user; For each actual fault category, if the actual fault category is one of the preset fault categories, the predicted fault probability corresponding to the actual fault category is determined as the actual fault probability; if the actual fault category is not one of the preset fault categories, the actual fault probability of the actual fault category is determined as the pre-configured probability. The actual fault category, the actual fault probability, and the feature vector are stored in the historical fault database as historical fault samples. Based on the matching between the actual fault category and each of the predicted fault categories, the deviation fault category is determined; For each of the aforementioned deviation fault categories, the difference between the actual fault probability and the predicted fault probability for that deviation fault category is calculated to obtain the deviation fault probability. If the probability of a deviation fault in at least one of the deviation fault categories is greater than a preset deviation threshold, the fault detection model is retrained using the historical fault samples.

7. The dishwasher according to claim 6, characterized in that, The controller is also configured to: For each of the preset fault categories, the actual fault probability of the preset fault category is extracted from the historical fault database, and the preset probability threshold corresponding to the preset fault category is adjusted according to the extracted actual fault probability.

8. The dishwasher according to claim 5, characterized in that, After all predicted fault categories have been repaired or a fault repair completion notification has been received from the user, the controller is further configured to: Obtain the actual fault category, actual fault root cause, and actual fault repair strategy input by the user; The waterway fault knowledge graph is updated based on the actual fault category, the actual fault root cause, and the actual fault repair strategy.

9. The dishwasher according to claim 1, characterized in that, The controller is also configured to: Based on the pressure signal at the inlet valve, determine the water inlet time required from opening to closing of the inlet valve, and calculate the remaining lifespan of the inlet valve based on the water inlet time. Based on the pressure signal at the drainage pump, the drainage time required from start-up to end of drainage is determined, and the remaining lifespan of the drainage pump is calculated based on the drainage time. The remaining lifespan of the filter is calculated based on the pressure difference signal before and after the filter. Based on the remaining lifespan of the inlet valve, the remaining lifespan of the drain pump, and the remaining lifespan of the filter, the corresponding warning levels for the inlet valve, the drain pump, and the filter are determined respectively, and corresponding warning prompt information is output according to the warning levels.

10. A method for detecting faults in a dishwasher, characterized in that, include: Acquire the sensing signals collected by each monitoring sensor group, wherein the sensing signals include: the current signal and pressure signal of the water inlet valve, the impedance signal associated with the water accumulation state at the bottom of the washing tank, the pressure difference signal before and after the filter and the turbidity signal of the circulating water, the rotation speed signal of the spray arm and the current signal of the circulating water pump, the current signal of the drain pump and the pressure signal at the drain pump. At least one statistical feature of the sensing signal is extracted, the extracted statistical features are combined into a feature vector, and the feature vector is input into the fault detection model to obtain the predicted fault probability of each preset fault category. Based on the predicted failure probability, the predicted failure category is determined; Based on the predicted fault category, the waterway fault knowledge graph is queried to obtain the predicted root cause of the waterway circuit fault.

Images