Fault diagnosis and maintenance system and method for intelligent camellia irrigation device

Abstract

The invention relates to the technical field of equipment fault diagnosis and maintenance, in particular to a fault diagnosis and maintenance system and method for an intelligent camellia irrigation device. Working state data of the camellia irrigation device is collected in real time through multiple sensors, noise is eliminated and data quality is optimized through a denoising and standardized data processing method, then abnormity is rapidly screened through threshold judgment, and the state of the device is accurately judged in combination with a convolutional neural network and a multi-layer decision fusion mechanism. According to the method, a fault classification model is utilized to finely divide fault types, finally, historical fault data comparison is carried out, fault roots and parts are speculated, specific maintenance suggestions are generated, finally, all information is displayed through a visual module, and real-time monitoring and decision support is provided for operation and maintenance personnel. The method effectively improves the operation reliability of equipment, reduces the downtime, reduces the maintenance cost, and prolongs the service life of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment fault diagnosis and maintenance, and in particular to a fault diagnosis and maintenance system and method for a camellia intelligent irrigation device. Background Art

[0002] With the continuous improvement of the level of agricultural intelligence, the demand for intelligent irrigation of high-value plants such as camellia is gradually increasing. Especially in the context of climate change and water shortage, the application of camellia intelligent irrigation devices is becoming more and more extensive. However, various faults may occur in the equipment during long-term operation, such as sensor damage, water flow blockage or electrical failure. These faults not only affect the irrigation effect, but may also cause long-term damage to the equipment and production stagnation. Traditional fault diagnosis methods rely on manual inspections, which are inefficient and cannot respond to changes in equipment status in real time. In addition, they fail to detect the fault in the early stage and often miss the best time for maintenance.

[0003] The Chinese invention patent with announcement number CN118037280B discloses a maintenance and fault diagnosis system for a corrugated paper production line, which includes the following modules: a sensor network module: deploying photoelectric sensors at each production node of the corrugated paper production line to monitor the actual operating speed of each production node; a speed matching monitoring module: receiving speed data collected by the sensor network module in real time, and using an intelligent algorithm to analyze whether the speeds of each process match; a fault diagnosis module: using a fault tree analysis method to diagnose specific faults or problem points that cause speed mismatches; a real-time maintenance module: generating a diagnostic report based on the results of the fault diagnosis module, and automatically adjusting the operating speed of related processes on the production line. The invention can automatically adjust production parameters to match speed requirements by continuously monitoring and analyzing the production speed of each node of the production line, and identify faulty production nodes by monitoring the production speed, while diagnosing the source of the fault.

[0004] In addition, existing fault diagnosis methods often ignore early warnings during equipment operation and fail to take effective measures in the early stages of a fault, resulting in aggravated faults and heavy maintenance work. Therefore, how to achieve real-time fault diagnosis of camellia intelligent irrigation devices, timely detect and effectively handle faults, has become a core issue that needs to be solved urgently. Summary of the invention

[0005] The purpose of the present invention is to address the problems existing in the background technology and to propose a fault diagnosis and maintenance system and method for a camellia intelligent irrigation device.

[0006] The technical solution of the present invention is a fault diagnosis and maintenance method for a camellia intelligent irrigation device, comprising the following specific steps:

[0007] S1. Using several sensors to collect working status data of the camellia intelligent irrigation device;

[0008] S2. Verify the work status data to ensure that the collected data is complete and valid, and avoid the impact of missing or abnormal values ​​on subsequent analysis. Then, based on the adaptive piecewise smoothing spatiotemporal data collaborative denoising method, denoise the work status data, and standardize the denoised work status data to obtain the processed real-time work status data.

[0009] S3. Obtain any real-time sensor data in the real-time working status data, introduce a threshold determination method, make a preliminary determination of the operating status based on the characteristic value, obtain a preliminary classification, and then input the real-time working status data into the pre-trained fault detection model, output the predicted status, and through a multi-layer decision fusion mechanism, combine the preliminary classification and predicted status of the current operating status, and determine the final status according to the fault priority principle;

[0010] If the final status of the current device is Normal, it means that the device has no faults and is in normal working condition;

[0011] If the final status of the current device is Warning or Fault, execute step S4;

[0012] Among them, Normal means that the device operates within the design parameter range and there is no sign of abnormality or failure; Warning means that some operating parameters of the device are close to the boundary of its normal range, but have not exceeded the safety limit; Fault means that the operating parameters of the device exceed the set safety range, resulting in impaired equipment function or obvious failure;

[0013] S4. If the final state of the current device is Warning, a continuous monitoring suggestion is generated; if the final state of the current device is Fault, the fault type is automatically identified and specific maintenance suggestions are generated;

[0014] S5. Display the working status data, the final status of the current device, the fault type and the maintenance suggestions on the visual interface, and generate a maintenance report for the camellia intelligent irrigation device.

[0015] Preferably, the denoising process for denoising the working status data is as follows:

[0016] S21. Calculate the rate of change of adjacent data points. The formula is:

[0017]

[0018] In the formula, R i represents the rate of change of the ith data point; D i and D i-1 Represent the i-th original data point and the i-1-th original data point respectively; D meanrepresents the data mean; N represents the total number of original data points;

[0019] S22, perform dynamic segmentation and adjust the segment size according to the change rate:

[0020]

[0021] Where, L i represents the length of the i-th segment; θ represents the set change rate threshold; L small and L large Respectively represent the preset length of the small segment and the preset length of the large segment;

[0022] S23, perform time dimension smoothing, and apply weighted moving average method to denoise based on segmentation, the formula is:

[0023]

[0024] In the formula, represents the i-th data point after denoising; w j Denotes weight; D j represents the jth original data point;

[0025] S24, perform spatial dimension correction, and correct the smoothed data in the time dimension according to the spatial correlation of the adjacent sensor data:

[0026]

[0027] In the formula, represents the i-th data point after spatial correction; α represents the balance coefficient between the time dimension and the smoothing dimension; S k represents the data of the kth related sensor at the corresponding moment; M represents the number of related sensors;

[0028] S25, the denoising result obtained in step S24 As the first round of estimates, a secondary correction is made based on the error between the original data and the estimated value:

[0029]

[0030] In the formula, represents the denoising result of the tth iteration; β represents the correction coefficient, which controls the correction amplitude;

[0031] The iteration stops when the preset condition is reached. The preset condition is:

[0032]

[0033] In the formula, ε represents the preset error threshold;

[0034] S26. Output the denoised working status data.

[0035] Preferably, the verification process for verifying the working status data is as follows:

[0036] S31. Use the missing rate formula to determine the degree of data missing:

[0037]

[0038] In the formula, the number of missing values ​​indicates the values ​​that the sensor has not successfully collected within a certain period of time; the total number of data points indicates the total number of data points that should be collected theoretically;

[0039] If the missing rate exceeds the set missing threshold, the data is marked invalid and a security warning is issued;

[0040] S32. Use the upper and lower check formulas to check the data validity to ensure that the collected working status data is within a reasonable range:

[0041]

[0042] Where D valid represents the output result of the finiteness test; D represents the collected original data points; L and U represent the lower and upper limits of the reasonable range of sensor data, respectively.

[0043] Preferably, the final state is generated as follows:

[0044] S41. Obtain real-time working status data i , get real-time working status data i Any real-time sensor data udata i ;

[0045] S42, introduce a threshold determination method, and make a preliminary determination of the operating state based on the characteristic value. The preliminary determination process is as follows:

[0046] S4201, extracting the normal range and warning range of any preset sensor data from the system database: warn,i , U warn,i , L normal,i , U normal,i};

[0047] Among them, L warn,i Indicates the lower limit threshold of the warning range of any sensor data; U warn,i Indicates the upper threshold of the warning range of any sensor data; L normal,i and U normal,i Respectively represent the upper and lower thresholds of the normal range of any sensor data; L warn,i <Lnormal,i ; U warn,i >U normal,i ;

[0048] S4202: Preliminary classification of the current operating state based on whether the characteristic value exceeds the above range:

[0049]

[0050] In the formula, S i Indicates the preliminary classification of the current operating status; Normal means that the device is operating within the design parameter range and there is no sign of abnormality or failure; Warning means that some operating parameters of the device are close to the boundary of its normal range, but have not exceeded the safety limit; Fault means that the operating parameters of the device exceed the set safety range, resulting in damage to the equipment function or obvious failure;

[0051] S43, real-time working status data i Input to the pre-trained fault detection model and output the predicted state

[0052] S44. Through the multi-layer decision fusion mechanism, the preliminary classification of the current operating status is integrated. i and predicted status The final status is determined according to the fault priority principle:

[0053]

[0054] in, Indicates the final status of the current device.

[0055] Preferably, the fault detection model is constructed based on a convolutional neural network, which extracts local features through a convolutional layer, reduces the dimension through a pooling layer, and finally performs preliminary classification through a fully connected layer.

[0056] Preferably, the generation process of the specific maintenance suggestion is as follows:

[0057] S61, if the final state of the current device is Warning, generating a continuous monitoring suggestion and transmitting it to the system visualization module;

[0058] If the final status of the current device is Fault, obtain the real-time working status data i Any real-time sensor data udata i , then implement step S62;

[0059] S62, real-time working status data i Input into the fault classification model, and output the type of output fault;

[0060] S63. Analyze possible faulty components based on the matching degree between historical fault data and current fault type. Calculate the similarity between historical fault characteristics and current fault characteristics. Select the historical fault with the closest similarity to the current abnormal mode as a reference. Infer the root cause of the fault and determine the location of the fault:

[0061]

[0062] In the formula, sim(F current ,F history ) indicates the current fault feature F current With historical fault characteristics F history similarity; Represents any real-time sensor data udata i The historical data of the corresponding sensor; n represents the number of data sources, that is, the number of sensors set;

[0063] S64. Calculate the severity score of the fault based on the location of the fault and its impact on the system:

[0064]

[0065] In the formula, I i represents the severity score of the fault; Im() represents the criticality of the current component to the operation of the device; Se() represents the degree of damage to the device caused by the current fault; α represents the weight coefficient, that is, the importance of the device; β represents the weight coefficient, that is, the severity of the fault;

[0066] S65. Automatically generate specific maintenance suggestions based on the type, root cause and severity of the fault.

[0067] Preferably, the fault classification model is constructed based on a support vector machine.

[0068] Preferably, the working status data includes water flow, soil moisture, temperature and pressure.

[0069] The technical solution of the present invention is a fault diagnosis and maintenance system for a camellia intelligent irrigation device, which is used to execute the fault diagnosis and maintenance method for the camellia intelligent irrigation device, including:

[0070] Data acquisition module, used to collect working status data of camellia intelligent irrigation device in real time;

[0071] The data processing module is used to process and analyze the collected sensor data and identify possible failure modes through a preset algorithm;

[0072] The fault diagnosis module is used to identify fault patterns based on the comparison of real-time data and historical data, and output fault diagnosis results using machine learning models;

[0073] Maintenance suggestion module, which is used to provide specific maintenance suggestions based on fault diagnosis results and push alarm information to users;

[0074] System visualization module, which provides a user-friendly graphical interface to facilitate users to view the real-time status, fault information and maintenance records of the irrigation system;

[0075] The system database is used to store data generated during the fault diagnosis and maintenance of the camellia intelligent irrigation device.

[0076] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0077] The present invention designs a fault diagnosis and maintenance system and method for a camellia intelligent irrigation device. By comprehensively using data processing, machine learning, and decision fusion technology, it ensures that the camellia intelligent irrigation device can achieve accurate fault diagnosis and efficient maintenance management, which not only improves the reliability and operation efficiency of the equipment, but also reduces the risk of failure and maintenance costs. It has significant technical advantages and practical application value:

[0078] (1) Improve the accuracy of fault detection: By performing multi-level verification, denoising, and standardization on the real-time data collected by the sensor, the interference of noise and outliers is effectively eliminated, providing a high-quality data foundation for subsequent fault detection. At the same time, the adaptive piecewise smoothing spatiotemporal data collaborative denoising method can dynamically adjust the denoising strategy according to the changes in the data, further improving the accuracy and reliability of fault detection;

[0079] (2) Rapidly respond to faults and optimize maintenance processes: By introducing threshold judgment, fault detection models, and decision fusion mechanisms, the system can quickly screen out abnormal data and accurately identify fault conditions. When a fault occurs, the system can detect it in a timely manner and give an accurate fault category. By comparing historical fault data and supporting fault classification models, the system can quickly determine the root cause of the fault and give targeted maintenance suggestions, thereby greatly shortening the response time for equipment maintenance.

[0080] (3) Enhance system intelligence and adaptability: By combining machine learning, decision fusion and adaptive algorithms, the system has a high degree of intelligence and adaptability. The combination of threshold judgment and machine learning model makes fault detection more accurate and robust, while the adaptive denoising method ensures that the system can automatically optimize its performance under different working environments, making the system more flexible in complex environments and having stronger fault prevention capabilities.

[0081] (4) Improve equipment stability and operating efficiency: Through real-time monitoring and intelligent fault diagnosis, the system can promptly detect and take measures when equipment abnormalities occur, thereby effectively avoiding long-term downtime and production interruptions caused by equipment failures. This mechanism not only improves equipment stability, but also reduces the impact of equipment failures on production, optimizes the overall operating efficiency of the equipment, and reduces maintenance costs;

[0082] (5) Accurately generate maintenance recommendations and optimize maintenance decisions: Based on the analysis of fault characteristics and comparison with historical fault patterns, the system can accurately identify the nature and scope of the fault, and generate corresponding maintenance recommendations based on the fault type, severity, and equipment importance, thereby helping operation and maintenance personnel to quickly locate the problem and take effective repair measures based on the nature and scope of the specific faulty component, thereby improving maintenance efficiency and accuracy and reducing the risks caused by improper maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 This is a system architecture diagram of a fault diagnosis and maintenance system for a camellia intelligent irrigation device proposed in the present invention. DETAILED DESCRIPTION

[0084] Embodiment 1, as Figure 1 As shown, a fault diagnosis and maintenance system for a camellia intelligent irrigation device proposed by the present invention includes: a data acquisition module, a data processing module, a fault diagnosis module, a maintenance suggestion module, a system visualization module and a system database.

[0085] The data acquisition module is used to collect the working status data of the camellia intelligent irrigation device in real time, including but not limited to water flow, soil moisture, temperature and pressure;

[0086] The data processing module is based on an embedded computing platform or a cloud computing platform, and is used to process and analyze the collected sensor data, and identify possible failure modes (including but not limited to abnormal flow, abnormal water pressure, and sensor failure) through a preset algorithm;

[0087] The fault diagnosis module is used to identify fault patterns based on the comparison of real-time data and historical data, and output fault diagnosis results using machine learning models;

[0088] The maintenance suggestion module is used to provide specific maintenance suggestions (including but not limited to equipment replacement, debugging methods, and repair steps) based on the fault diagnosis results, and can push alarm information to users;

[0089] The system visualization module is used to provide a user-friendly graphical interface, allowing users to view the real-time status, fault information and maintenance records of the irrigation system;

[0090] The system database is used to store data generated during the fault diagnosis and maintenance of the camellia intelligent irrigation device.

[0091] Embodiment 2, a fault diagnosis and maintenance method of a camellia intelligent irrigation device proposed in the present invention is applied to a fault diagnosis and maintenance system of a camellia intelligent irrigation device proposed in Embodiment 1, and its specific implementation steps are as follows:

[0092] S1. After the system is started, the data acquisition module instructs the configured sensor units to enter the working state, collects the working state data of the camellia intelligent irrigation device, and then transmits the collected working state data to the data processing module;

[0093] The working status data includes but is not limited to water flow, soil moisture, temperature and pressure.

[0094] S2. The data processing module receives the working status data and processes the working status data. The processing process is as follows:

[0095] S21. Verify the working status data to ensure that the collected data is complete and valid, and avoid the impact of missing or abnormal values ​​on subsequent analysis. The verification process is as follows:

[0096] S2101. Use the missing rate formula to determine the degree of data missingness:

[0097]

[0098] In the formula, the number of missing values ​​indicates the values ​​that the sensor has not successfully collected within a certain period of time; the total number of data points indicates the total number of data points that should be collected theoretically;

[0099] If the missing rate exceeds the set missing threshold (set to 10% in this embodiment), the data is marked as invalid and a security warning is issued;

[0100] S2102. Use the upper and lower check formulas to check data validity to ensure that the collected working status data is within a reasonable range:

[0101]

[0102] Where D valid represents the output result of the finiteness test; D represents the collected original data points; L and U represent the lower and upper limits of the reasonable range of sensor data, respectively;

[0103] S22. Based on the adaptive piecewise smoothing spatiotemporal data collaborative denoising method, the working status data is denoised to eliminate random fluctuations and noise in the data:

[0104] S2201, calculate the change rate of adjacent data points, the formula is:

[0105]

[0106] In the formula, R i represents the rate of change of the ith data point; D i and D i-1 Represent the i-th original data point and the i-1-th original data point respectively; D mean represents the data mean; N represents the total number of original data points;

[0107] S2202, perform dynamic segmentation and adjust the segment size according to the change rate:

[0108]

[0109] Where, L i represents the length of the i-th segment; θ represents the set change rate threshold; L small and L large Respectively represent the preset length of the small segment and the preset length of the large segment;

[0110] S2203, perform time dimension smoothing processing, and apply weighted moving average method to denoise based on segmentation, the formula is:

[0111]

[0112] In the formula, represents the i-th data point after denoising; w j represents the weight, and in this embodiment, it is set to be a Gaussian distribution weight with a higher center weight; D j represents the jth original data point;

[0113] S2204: Perform spatial dimension correction, and correct the smoothed data in the time dimension according to the spatial correlation of the adjacent sensor data:

[0114]

[0115] In the formula, represents the i-th data point after spatial correction; α represents the balance coefficient between the time dimension and the smoothing dimension, and the embodiment sets its range to [0.5, 0.8]; S k represents the data of the kth related sensor at the corresponding moment; M represents the number of related sensors;

[0116] S2205: De-noising the result obtained in step S2204 As the first round of estimates, a secondary correction is made based on the error between the original data and the estimated value:

[0117]

[0118] In the formula, represents the denoising result of the tth iteration; β represents the correction coefficient, which controls the correction amplitude;

[0119] The iteration stops when the preset condition is reached. The preset condition is:

[0120]

[0121] In the formula, ε represents the preset error threshold;

[0122] S2206, outputting the denoised working status data data;

[0123] S23, standardize the denoised working status data, and the standardization formula is:

[0124]

[0125] In the formula, datas represents the data after standardization; data i represents the i-th data point after denoising; N represents the total number of data points;

[0126] S24, the processed real-time working status data, that is, the real-time standardized processed data datas i Transmitted to the fault diagnosis module.

[0127] S3, the fault diagnosis module, based on data preprocessing, ensures that the system can accurately detect abnormal conditions in equipment operation through the collaboration of feature extraction, threshold determination, multi-layer decision fusion and machine learning classification. The specific implementation process is as follows:

[0128] S31. Obtain real-time working status data i , get real-time working status data i Any real-time sensor data udata i ;

[0129] S32. In order to quickly screen abnormal data and reduce the calculation burden of complex models, the system introduces a threshold judgment method to make a preliminary judgment on the operating status based on the characteristic value. The preliminary judgment process is as follows:

[0130] S3201, extracting the normal range and warning range of any preset sensor data from the system database: warn,i , U warn,i , L normal,i , U normal,i};

[0131] Among them, L warn,iIndicates the lower limit threshold of the warning range of any sensor data; U warn,i Indicates the upper threshold of the warning range of any sensor data; L normal,i and U normal,i Respectively represent the upper and lower thresholds of the normal range of any sensor data; L warn,i <L normal,i ; U warn,i >U normal,i ;

[0132] S3202: Preliminary classification of the current operating state based on whether the characteristic value exceeds the above range:

[0133]

[0134] In the formula, S i Indicates the preliminary classification of the current operating status; Normal means that the device is operating within the design parameter range and there is no sign of abnormality or failure; Warning means that some operating parameters of the device are close to the boundary of its normal range, but have not exceeded the safety limit; Fault means that the operating parameters of the device exceed the set safety range, resulting in damage to the equipment function or obvious failure;

[0135] S33, real-time working status data i Input to the pre-trained fault detection model and output the predicted state

[0136] It should be noted that the fault detection model is built based on a convolutional neural network, which extracts local features through the convolution layer, reduces the dimension through the pooling layer, and finally performs classification through the fully connected layer;

[0137] S34. Through the multi-layer decision fusion mechanism, the preliminary classification of the current operating status is integrated. i and predicted status The final status is determined according to the fault priority principle:

[0138]

[0139] in, Indicates the final state of the current device;

[0140] S35, if the final state of the current device Normal, the real-time working status data i and the final state of the current device Transmit to the system visualization module and save to the system database;

[0141] If the final state of the current device Warning or Fault, the final status of the current device Transmitted to the maintenance suggestion module and saved to the system database.

[0142] S4, the maintenance suggestion module further analyzes the cause, nature and impact of the fault based on real-time fault detection, and identifies the fault type, faulty components and production maintenance suggestions. The specific implementation process is as follows:

[0143] S41, if the final state of the current device When it is Warning, continuous monitoring suggestions are generated and transmitted to the system visualization module;

[0144] If the final state of the current device When it is Fault, get real-time working status data i , get datas i Any real-time sensor data udata i , then implement step S42;

[0145] S42, real-time working status data i Input into the fault classification model, and output the type of output fault

[0146] S43. Analyze possible faulty components based on the matching degree between historical fault data and current fault type. Calculate the similarity between historical fault characteristics and current fault characteristics. Select the historical fault closest to the current abnormal mode as a reference. Infer the root cause of the fault and determine the location of the fault:

[0147]

[0148] In the formula, sim(F current ,F history ) indicates the current fault feature F current With historical fault characteristics F history similarity; Represents any real-time sensor data udata i The historical data of the corresponding sensor; n represents the number of data sources, that is, the number of sensors set;

[0149] It should be noted that the fault classification model is built on the basis of support vector machine (SVM). The goal of support vector machine (SVM) is to find a hyperplane that can classify samples in the data set into different categories. Support vector machine achieves the best segmentation by maximizing the margin, thereby improving the accuracy of classification. For nonlinear problems, SVM maps data to a higher-dimensional space by introducing a kernel function to achieve linear separability.

[0150] S44. Calculate the severity score of the fault based on the location of the fault and its impact on the system:

[0151]

[0152] In the formula, I i represents the severity score of the fault; Im() represents the criticality of the current component to the operation of the device; Se() represents the degree of damage to the device caused by the current fault; α represents the weight coefficient, that is, the importance of the device; β represents the weight coefficient, that is, the severity of the fault;

[0153] S45, automatically generate specific maintenance suggestions based on the type, root cause and severity of the fault;

[0154] For example, if it is a sensor failure, it can be recommended to replace the sensor; if it is an electrical failure, it can be recommended to check the circuit board or replace the power module;

[0155] S46. Transmit the fault type and the generated maintenance suggestion to the system visualization module.

[0156] S5. The system visualization module displays the implementation working status data and the final status of the current device on the visualization interface. Fault type and maintenance suggestions, and generate a camellia intelligent irrigation device maintenance report, which will then be stored in the system database.

[0157] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A fault diagnosis and maintenance method for a camellia intelligent irrigation device, characterized in that: The specific steps include: S1. Using several sensors to collect working status data of the camellia intelligent irrigation device; S2. Verify the work status data to ensure that the collected data is complete and valid, and avoid the impact of missing or abnormal values ​​on subsequent analysis. Then, based on the adaptive piecewise smoothing spatiotemporal data collaborative denoising method, denoise the work status data, and standardize the denoised work status data to obtain the processed real-time work status data. S3. Obtain any real-time sensor data in the real-time working status data, introduce a threshold determination method, make a preliminary determination of the operating status based on the characteristic value, obtain a preliminary classification, and then input the real-time working status data into the pre-trained fault detection model, output the predicted status, and through a multi-layer decision fusion mechanism, combine the preliminary classification and predicted status of the current operating status, and determine the final status according to the fault priority principle; If the final status of the current device is Normal, it means that the device has no faults and is in normal working condition; If the final status of the current device is Warning or Fault, execute step S4; Among them, Normal means that the device operates within the design parameter range and there is no sign of abnormality or failure; Warning means that some operating parameters of the device are close to the boundary of its normal range, but have not exceeded the safety limit; Fault means that the operating parameters of the device exceed the set safety range, resulting in impaired equipment function or obvious failure; S4. If the final state of the current device is Warning, a continuous monitoring suggestion is generated; if the final state of the current device is Fault, the fault type is automatically identified and specific maintenance suggestions are generated; S5. Display the working status data, the final status of the current device, the fault type and the maintenance suggestions on the visual interface, and generate a maintenance report for the camellia intelligent irrigation device.

2. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: The denoising process for working status data is as follows: S21. Calculate the rate of change of adjacent data points. The formula is: In the formula, R i represents the rate of change of the ith data point; D i and D i-1 Represent the i-th original data point and the i-1-th original data point respectively; D mean represents the data mean; N represents the total number of original data points; S22, perform dynamic segmentation and adjust the segment size according to the change rate: Where, L i represents the length of the i-th segment; θ represents the set change rate threshold; L small and L large Respectively represent the preset length of the small segment and the preset length of the large segment; S23, perform time dimension smoothing, and apply weighted moving average method to denoise based on segmentation, the formula is: In the formula, represents the i-th data point after denoising; w j Denotes weight; D j represents the jth original data point; S24, perform spatial dimension correction, and correct the smoothed data in the time dimension according to the spatial correlation of the adjacent sensor data: In the formula, represents the i-th data point after spatial correction; α represents the balance coefficient between the time dimension and the smoothing dimension; S k represents the data of the kth related sensor at the corresponding moment; M represents the number of related sensors; S25, the denoising result obtained in step S24 As the first round of estimates, a secondary correction is made based on the error between the original data and the estimated value: In the formula, represents the denoising result of the tth iteration; β represents the correction coefficient, which controls the correction amplitude; The iteration stops when the preset condition is reached. The preset condition is: In the formula, ε represents the preset error threshold; S26. Output the denoised working status data.

3. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: The verification process for working status data is as follows: S31. Use the missing rate formula to determine the degree of data missing: In the formula, the number of missing values ​​indicates the values ​​that the sensor has not successfully collected within a certain period of time; the total number of data points indicates the total number of data points that should be collected theoretically; If the missing rate exceeds the set missing threshold, the data is marked invalid and a security warning is issued; S32. Use the upper and lower check formulas to check the data validity to ensure that the collected working status data is within a reasonable range: Where D valid represents the output result of the finiteness test; D represents the collected original data points; L and U represent the lower and upper limits of the reasonable range of sensor data, respectively.

4. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: The final state is generated as follows: S41. Obtain real-time working status data i , get real-time working status data i Any real-time sensor data udata i ; S42, introduce a threshold determination method, and make a preliminary determination of the operating state based on the characteristic value. The preliminary determination process is as follows: S4201, extracting the normal range and warning range of any preset sensor data from the system database: warn,i , U warn,i , L normal,i , U normal,i }; Among them, L warn,i Indicates the lower limit threshold of the warning range of any sensor data; U warn,i Indicates the upper threshold of the warning range of any sensor data; L normal,i and U normal,i Respectively represent the upper and lower thresholds of the normal range of any sensor data; L warn,i <L normal,i ; U warn,i >U normal,i ; S4202: Preliminary classification of the current operating state based on whether the characteristic value exceeds the above range: In the formula, S i Indicates the preliminary classification of the current operating status; Normal means that the device is operating within the design parameter range and there is no sign of abnormality or failure; Warning means that some operating parameters of the device are close to the boundary of its normal range, but have not exceeded the safety limit; Fault means that the operating parameters of the device exceed the set safety range, resulting in damage to the equipment function or obvious failure; S43, real-time working status data i Input to the pre-trained fault detection model and output the predicted state S44. Through the multi-layer decision fusion mechanism, the preliminary classification of the current operating status is integrated. i and predicted status The final status is determined according to the fault priority principle: in, Indicates the final status of the current device.

5. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 4, characterized in that: The fault detection model is built based on a convolutional neural network. The local features are extracted through the convolution layer, the dimension is reduced through the pooling layer, and finally the preliminary classification is performed through the fully connected layer.

6. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: The specific maintenance suggestion generation process is as follows: S61, if the final state of the current device is Warning, generating a continuous monitoring suggestion and transmitting it to the system visualization module; If the final status of the current device is Fault, obtain the real-time working status data i Any real-time sensor data udata i , then implement step S62; S62, real-time working status data i Input into the fault classification model, and output the type of output fault; S63. Analyze possible faulty components based on the matching degree between historical fault data and current fault type. Calculate the similarity between historical fault characteristics and current fault characteristics. Select the historical fault with the closest similarity to the current abnormal mode as a reference. Infer the root cause of the fault and determine the location of the fault: In the formula, sim(F current ,F history ) indicates the current fault feature F current With historical fault characteristics F history similarity; Represents any real-time sensor data udata i The historical data of the corresponding sensor; n represents the number of data sources, that is, the number of sensors set; S64. Calculate the severity score of the fault based on the location of the fault and its impact on the system: In the formula, I i represents the severity score of the fault; Im() represents the criticality of the current component to the operation of the device; Se() represents the degree of damage to the device caused by the current fault; α represents the weight coefficient, that is, the importance of the device; β represents the weight coefficient, that is, the severity of the fault; S65. Automatically generate specific maintenance suggestions based on the type, root cause and severity of the fault.

7. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: The fault classification model is built based on support vector machine.

8. The fault diagnosis and maintenance method of a camellia intelligent irrigation device according to claim 1, characterized in that: Working status data includes water flow, soil moisture, temperature and pressure.

9. A fault diagnosis and maintenance system for a camellia intelligent irrigation device, which is used to execute a fault diagnosis and maintenance method for a camellia intelligent irrigation device according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, used to collect working status data of camellia intelligent irrigation device in real time; The data processing module is used to process and analyze the collected sensor data and identify possible failure modes through a preset algorithm; The fault diagnosis module is used to identify fault patterns based on the comparison of real-time data and historical data, and output fault diagnosis results using machine learning models; Maintenance suggestion module, which is used to provide specific maintenance suggestions based on fault diagnosis results and push alarm information to users; System visualization module, which provides a user-friendly graphical interface to facilitate users to view the real-time status, fault information and maintenance records of the irrigation system; The system database is used to store data generated during the fault diagnosis and maintenance of the camellia intelligent irrigation device.

Images