Laboratory fault automatic diagnosis system and method
By collecting laboratory data in real time through sensors and machine vision modules, and combining them with data processing and analysis modules for fault diagnosis, the problem of laboratory fault diagnosis relying on manual experience is solved, and fast and accurate fault identification and repair suggestions are achieved, ensuring stable operation of the laboratory.
Patent Information
- Application Number
- CN202510563779.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-30
- Publication Date
- 2025-09-12
AI Technical Summary
Existing laboratory fault diagnosis relies on manual experience, which is tedious and time-consuming, resulting in inaccurate diagnosis, affecting laboratory operations and the accuracy of experimental data, and may cause greater losses.
The sensor module and machine vision module are used to collect laboratory data in real time, and the data processing and analysis module is combined to perform preprocessing and machine learning model training to identify the fault type and location, and generate fault reports and repair suggestions.
It achieves fast and accurate fault diagnosis, reduces labor costs, reduces maintenance costs and experimental interruption costs caused by human errors, improves fault discovery and response speed, and ensures safe and stable operation of the laboratory.
Smart Images
Figure CN120630930A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a field of laboratory fault automatic diagnosis system and method. Background Art
[0002] With the continuous advancement of science and technology, laboratory equipment and facilities are becoming increasingly complex, involving a growing number of sub-disciplines. For example, laboratories in fields such as biology, chemistry, and physics require complex instrumentation and supporting environments. Laboratory equipment typically includes a variety of precision instruments, sensors, control systems, and supporting facilities (such as plumbing, air conditioning, and gas supply systems). When a laboratory malfunction occurs, due to the high level of expertise and the complexity of troubleshooting, general maintenance personnel (such as plumbers) often struggle to quickly and accurately diagnose the problem, resulting in laboratory downtime, impacting the accuracy of experimental data and the progress of experiments, and potentially even causing significant losses.
[0003] Traditional fault diagnosis usually relies on manual experience, and the maintenance process is cumbersome, which not only consumes a lot of time but may also cause unnecessary losses due to human errors. Therefore, there is an urgent need for an automated and intelligent fault diagnosis and repair suggestion system that can monitor various equipment and systems in the laboratory in real time and provide efficient and accurate diagnosis results and solutions when faults occur, thereby minimizing laboratory downtime and ensuring normal laboratory operations. Summary of the Invention
[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a laboratory fault automatic diagnosis system and method.
[0005] To achieve the above object, the present invention adopts the following technical solutions: A laboratory fault automatic diagnosis system and method, the laboratory fault automatic diagnosis system includes a sensor module, a machine vision module, a data processing and analysis module, and a fault reporting and suggestion module; The sensor module is used to collect laboratory environment data in real time; The machine vision module is used to collect images or videos of equipment in the laboratory in real time; The data processing and analysis module is used to process and analyze sensor data, device images or videos in real time to diagnose the specific fault type and fault location; The fault reporting and suggestion module is used to generate a fault report based on the diagnosed fault type and fault location, and provide preliminary solution suggestions.
[0006] Specifically, the laboratory fault automatic diagnosis method includes the following steps: S1: collect laboratory environment data in real time; The laboratory is equipped with cameras and various sensors; The sensors include air flow sensors, temperature and humidity sensors, gas concentration sensors, pressure sensors, cleanliness sensors, etc. The sensor module collects laboratory environmental data in real time through sensors, including air volume, temperature, humidity, odor concentration, pressure, concentration of particulate matter in the air, etc. Specifically, the air volume of the laboratory exhaust equipment is collected in real time through the air flow sensor; The temperature and humidity of the experiment are collected in real time through temperature and humidity sensors; The odor concentration in the laboratory is collected in real time through a gas concentration sensor; The laboratory pressure is collected in real time through the pressure sensor; The concentration of particulate matter in the laboratory air is collected in real time through cleanliness sensors; S2: Real-time monitoring of equipment status in the laboratory; The sensors installed in the laboratory also include infrared sensors, etc. The machine vision module collects real-time images or videos of laboratory equipment through cameras and infrared sensors; S3: Analyze and process the collected data; The sensor module and the machine vision module respectively transmit the collected real-time laboratory environment data and laboratory equipment images or videos to the data analysis and processing module. The data analysis and processing module receives the data and processes it, including the following sub-steps: S31: preprocessing data; The data analysis and processing module pre-processes the received laboratory environment data and laboratory equipment images or videos; The preprocessing includes processing missing data, outliers, etc.; Specifically, for the received laboratory environment data, the data analysis and processing module uses the pandas library to identify whether there are missing values. If so, the missing values in the laboratory environment data are filled by interpolation. Use statistical methods to identify whether there are outliers in the laboratory environment data, such as the Z-Score method and the box plot method. If there are outliers, replace the outliers with values within the mean or standard deviation range or delete the outliers directly; Specifically, the formula for detecting outliers using the Z-Score method is as follows: ; Where X is a specific data value in the collected environmental data, μ is the mean of all such data collected, and σ is the standard deviation of all such data collected. If the obtained Z value is greater than 3 or less than -3, the data is considered to be an outlier. For the real-time images or videos received from laboratory equipment, image processing techniques, including Gaussian blur and mean filtering, are used to remove image noise caused by poor lighting or sensor problems. The image pixel values are normalized by scaling them to the range of [0, 1]. The video is processed frame by frame to remove background and redundant information. The resulting data is saved in the laboratory dataset. S32: perform fault diagnosis; The data processing and analysis model pre-trains the machine learning model and performs real-time fault diagnosis on the data in the laboratory dataset based on the trained machine learning model. The specific steps include the following: S321: Collect laboratory historical failure data; Collect laboratory historical failure data from a laboratory historical failure database; The laboratory historical fault data includes fault type, fault occurrence time, fault description, fault cause, fault location, laboratory environment data, laboratory equipment information, maintenance records, etc. The fault types include environmental faults, equipment faults, etc. Environmental failures include no air flow or too low air volume in the laboratory exhaust equipment, excessive room odor, excessive laboratory temperature and humidity, excessive laboratory cleanliness, and laboratory pressure imbalance; equipment failures include control system failures and mechanical failures; The laboratory environment data includes air volume, temperature, humidity, odor concentration, pressure, concentration of particulate matter in the air, etc.; the laboratory equipment information includes equipment model, usage status, images and videos of equipment failure, etc.; The laboratory historical failure database contains various historical failure data of the laboratory; Preprocess the collected laboratory historical failure data; S322: performing feature extraction; For laboratory environmental data, time series analysis methods are used to extract features, including the mean and standard deviation of temperature, humidity, pressure, and other data; For laboratory equipment information, the image feature extraction method is used to extract edge features, color features, texture features, etc. from the images of equipment failure. A 3D convolutional neural network method is used to extract the spatiotemporal features of equipment failure from the video of equipment failure. The extracted features and the corresponding fault type and fault location are saved as a fault sample data in a data set, wherein the data set contains multiple fault sample data; S323: Model training; The dataset is divided into a training set and a validation set in proportion. The training set is input into the machine learning model. The supervised learning method is used to train the model by taking the fault type and fault location in each fault sample data as labels and the corresponding features as inputs. Furthermore, the features in each fault sample data are associated with the corresponding fault type and fault location; After the training is completed, the validation set is used to evaluate the model and optimize the machine learning model parameters until the machine learning model parameters are trained and the expected effect is achieved, thus obtaining a trained machine learning model. S324: Perform fault diagnosis; The data processing and analysis module inputs the laboratory environment data and laboratory equipment information in the laboratory data set obtained in step S31 into the fault diagnosis model in real time. The machine learning model makes a judgment based on the learned features, fault type, and location. If the machine learning model identifies that there is no fault, the fault diagnosis result output is "no fault"; If the machine learning model identifies a fault, the fault diagnosis result output is the fault type and fault location; S4: Generate a fault report; The data processing and analysis module transmits the fault diagnosis results to the fault reporting and suggestion module, which receives the data and generates a fault report based on the diagnosed fault type and fault location; The fault report includes the fault time, corresponding laboratory environment data and laboratory equipment data, fault type and location, fault cause analysis, preliminary repair suggestions, preventive measures, etc. Furthermore, the preliminary repair suggestion includes repair time and repair cost; If there are multiple preliminary repair suggestions, the optimal repair suggestion is recommended through a linear programming model, the content of which is as follows: ; Among them, the range of i is [1, n], which means there are n preliminary repair suggestions, which are 、 ... ; The value of is 0 or 1, =1 means select this repair suggestion, =0 means not to select the repair suggestion; the corresponding n repair times are 、 ... , the corresponding n repair costs are 、 ... ; and set the maximum available repair time to , the maximum available repair cost is ; The constraints are as follows: Only one repair suggestion can be selected for each fault; ; ; Based on this linear programming model, the optimal initial repair proposal is recommended by comprehensively considering factors such as repair time and cost; The fault report will be notified to relevant staff via email, text message, etc.
[0007] Compared with the prior art, the present invention has the following beneficial effects: The laboratory fault automatic diagnosis system and method proposed in the present invention diagnose the laboratory environment and equipment by integrating machine learning algorithms and data processing technologies, identify the fault type and location, and do not need to rely on a large number of manual inspections and fault judgments, thereby reducing labor costs, avoiding maintenance costs and experimental interruption costs caused by human judgment errors, greatly improving the speed of fault discovery and response, and reducing delays caused by human negligence.
[0008] At the same time, this method can monitor various environmental parameters in the laboratory in real time and detect anomalies in a timely manner, which can effectively avoid safety hazards caused by environmental anomalies; by collecting historical laboratory failure data over a long period of time, the system can discover potential failure trends and provide preventive maintenance recommendations, thereby reducing the failure rate of equipment and improving the operating stability of laboratory equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 The present invention is a flowchart of the steps of a method for automatic diagnosis of laboratory faults. DETAILED DESCRIPTION
[0010] In order to provide a further understanding of the purpose, structure, features, and functions of the present invention, the present invention is described in detail below with reference to the embodiments.
[0011] To achieve the above object, the present invention adopts the following technical solutions: like Figure 1 As shown, a laboratory fault automatic diagnosis system and method, the laboratory fault automatic diagnosis system includes a sensor module, a machine vision module, a data processing and analysis module, and a fault reporting and suggestion module; The sensor module is used to collect laboratory environment data in real time; The machine vision module is used to collect images or videos of equipment in the laboratory in real time; The data processing and analysis module is used to process and analyze sensor data, device images or videos in real time to diagnose the specific fault type and fault location; The fault reporting and suggestion module is used to generate a fault report based on the diagnosed fault type and fault location, and provide preliminary solution suggestions.
[0012] Specifically, the laboratory fault automatic diagnosis method includes the following steps: S1: collect laboratory environment data in real time; The laboratory is equipped with cameras and various sensors; The sensors include air flow sensors, temperature and humidity sensors, gas concentration sensors, pressure sensors, cleanliness sensors, etc. The sensor module collects laboratory environmental data in real time through sensors, including air volume, temperature, humidity, odor concentration, pressure, concentration of particulate matter in the air, etc. Specifically, the air volume of the laboratory exhaust equipment is collected in real time through the air flow sensor; The temperature and humidity of the experiment are collected in real time through temperature and humidity sensors; The odor concentration in the laboratory is collected in real time through a gas concentration sensor; The laboratory pressure is collected in real time through the pressure sensor; The cleanliness sensor collects the concentration of particulate matter in the laboratory air in real time.
[0013] S2: Real-time monitoring of equipment status in the laboratory; The sensors installed in the laboratory also include infrared sensors, etc. The machine vision module collects real-time images or videos of laboratory equipment through cameras and infrared sensors.
[0014] This method combines sensor modules and machine vision modules to collect environmental data (such as temperature, humidity, gas concentration, etc.) and equipment status information through images or videos. This combination of multimodal data can comprehensively and accurately reflect the fault conditions of laboratory equipment.
[0015] S3: Analyze and process the collected data; The sensor module and the machine vision module respectively transmit the collected real-time laboratory environment data and laboratory equipment images or videos to the data analysis and processing module. The data analysis and processing module receives the data and processes it, including the following sub-steps: S31: preprocessing data; The data analysis and processing module pre-processes the received laboratory environment data and laboratory equipment images or videos; The preprocessing includes processing missing data, outliers, etc.; Specifically, for the received laboratory environment data, the data analysis and processing module uses the pandas library to identify whether there are missing values. If so, the missing values in the laboratory environment data are filled by interpolation. Use statistical methods to identify whether there are outliers in the laboratory environment data, such as the Z-Score method and the box plot method. If there are outliers, replace the outliers with values within the mean or standard deviation range or delete the outliers directly; For the real-time images or videos received from laboratory equipment, image processing techniques, including Gaussian blur and mean filtering, are used to remove image noise caused by poor lighting or sensor problems. The image pixel values are normalized by scaling them to the range of [0, 1]. The video is processed frame by frame to remove background and redundant information. The resulting data is saved in the laboratory dataset. By preprocessing the data, the quality and reliability of the data are ensured while helping to improve the accuracy of image recognition; S32: perform fault diagnosis; The data processing and analysis model pre-trains the machine learning model and performs real-time fault diagnosis on the data in the laboratory dataset based on the trained machine learning model. The specific steps include the following: S321: Collect laboratory historical failure data; Collect laboratory historical failure data from a laboratory historical failure database; The laboratory historical fault data includes fault type, fault occurrence time, fault description, fault cause, fault location, laboratory environment data, laboratory equipment information, maintenance records, etc. The fault types include environmental faults, equipment faults, etc. Environmental failures include no air flow or too low air volume in the laboratory exhaust equipment, excessive room odor, excessive laboratory temperature and humidity, excessive laboratory cleanliness, and laboratory pressure imbalance; equipment failures include control system failures and mechanical failures; The laboratory environment data includes air volume, temperature, humidity, odor concentration, pressure, concentration of particulate matter in the air, etc.; the laboratory equipment information includes equipment model, usage status, images and videos of equipment failure, etc.; The laboratory historical failure database contains various historical failure data of the laboratory; Preprocess the collected laboratory historical failure data; S322: performing feature extraction; For laboratory environmental data, time series analysis methods are used to extract features, including the mean and standard deviation of temperature, humidity, pressure, and other data; For laboratory equipment information, the image feature extraction method is used to extract edge features, color features, texture features, etc. from the images of equipment failure. Edges in images are important indicators of equipment failure. Equipment failure may manifest as color changes, especially when the equipment is overheating or leaking gas. Color features can be used to determine whether the equipment is faulty. Use methods such as gray-level co-occurrence matrix (GLCM) to extract image texture information to reflect whether there is damage or abnormality on the device surface; A 3D convolutional neural network method is used to extract the spatiotemporal features of equipment failure from the video of equipment failure. The spatiotemporal features can capture the action patterns and abnormal behaviors during the fault process, such as vibration of the equipment and changes in heat maps; The extracted features and the corresponding fault type and fault location are saved as a fault sample data in a data set, wherein the data set contains multiple fault sample data; S323: Model training; The dataset is divided into a training set and a validation set in proportion. The training set is input into the machine learning model. The supervised learning method is used to train the model by taking the fault type and fault location in each fault sample data as labels and the corresponding features as inputs. Furthermore, the features in each fault sample data are associated with the corresponding fault type and fault location; After the training is completed, the validation set is used to evaluate the model and optimize the machine learning model parameters until the machine learning model parameters are trained and the expected effect is achieved, thus obtaining a trained machine learning model. Using machine learning models and image processing technology, this method can perform in-depth analysis of sensor data, equipment images or videos. By learning from historical fault data, the system can identify and predict the type and location of possible equipment faults. Feature extraction uses advanced technologies such as time series analysis, image feature extraction and 3D convolutional neural networks, making fault diagnosis not only accurate, but also able to predict and correct based on real-time data.
[0016] S324: Perform fault diagnosis; The data processing and analysis module inputs the laboratory environment data and laboratory equipment information in the laboratory data set obtained in step S31 into the fault diagnosis model in real time. The machine learning model makes a judgment based on the learned features, fault type, and location. If the machine learning model identifies that there is no fault, the fault diagnosis result output is "no fault"; If the machine learning model identifies a fault, the fault diagnosis result output is the fault type and fault location; S4: Generate a fault report; The data processing and analysis module transmits the fault diagnosis results to the fault reporting and suggestion module, which receives the data and generates a fault report based on the diagnosed fault type and fault location; The fault report includes the fault time, corresponding laboratory environment data and laboratory equipment data, fault type and location, fault cause analysis, preliminary repair suggestions, preventive measures, etc. This method provides preliminary solution suggestions and preventive measures based on the fault type and location, which provides clear guidance for laboratory staff to help them take quick measures to repair the fault and reduce the losses caused by the fault.
[0017] The fault report will be notified to relevant staff via email, text message, etc.
[0018] The present invention has been described with reference to the above embodiments. However, the above embodiments are merely exemplary embodiments of the present invention. It should be noted that the disclosed embodiments do not limit the scope of the present invention. On the contrary, modifications and improvements that do not depart from the spirit and scope of the present invention are intended to be protected by the present invention.
Claims
1. A method for automatic diagnosis of laboratory faults, characterized by: The following steps are involved: S1: collect laboratory environment data in real time; The laboratory is equipped with cameras and various sensors; the sensor module collects laboratory environment data in real time through sensors; S2: Real-time monitoring of equipment status in the laboratory; The machine vision module collects real-time images or videos of laboratory equipment through cameras and infrared sensors; S3: Analyze and process the collected data; The sensor module and the machine vision module respectively transmit the collected real-time laboratory environment data and laboratory equipment images or videos to the data analysis and processing module. The data analysis and processing module receives the data and processes it, including the following sub-steps: S31: preprocessing data; The data analysis and processing module pre-processes the received laboratory environment data and laboratory equipment images or videos; S32: perform fault diagnosis; In the data processing and analysis model, the machine learning model is trained in advance and the fault diagnosis of the data in the laboratory data set is performed in real time based on the trained machine learning model; S4: Generate a fault report; The data processing and analysis module transmits the fault diagnosis results to the fault reporting and suggestion module. The fault reporting and suggestion module receives the data and generates a fault report based on the diagnosed fault type and fault location.
2. The automatic laboratory fault diagnosis method according to claim 1, wherein: In step S1, the sensors include an air flow sensor, a temperature and humidity sensor, a gas concentration sensor, a pressure sensor, a cleanliness sensor, and an infrared sensor; The laboratory environmental data include air volume, temperature, humidity, odor concentration, pressure, and concentration of particulate matter in the air; Specifically, the air volume of the laboratory exhaust equipment is collected in real time through the air flow sensor; The temperature and humidity of the experiment are collected in real time through temperature and humidity sensors; The odor concentration in the laboratory is collected in real time through a gas concentration sensor; The laboratory pressure is collected in real time through the pressure sensor; The cleanliness sensor collects the concentration of particulate matter in the laboratory air in real time.
3. The automatic laboratory fault diagnosis method according to claim 1, wherein: The specific contents of step S31 are as follows: The preprocessing includes processing missing data and outliers; Specifically, for the received laboratory environment data, the data analysis and processing module uses the pandas library to identify whether there are missing values. If so, the missing values in the laboratory environment data are filled by interpolation. Use statistical methods to identify whether there are outliers in laboratory environmental data, and if so, replace the outliers with values within the mean, standard deviation, or delete them directly; For the real-time images or videos received from laboratory equipment, image processing techniques, including Gaussian blur and mean filtering, are used to remove image noise caused by poor lighting or sensor problems; the pixel values of the image are normalized by scaling them to the range of [0, 1]; the video is processed frame by frame, and the obtained data is saved in the laboratory dataset.
4. The automatic laboratory fault diagnosis method according to claim 1, wherein: The specific contents of step S32 are as follows: S321: Collect laboratory historical failure data; Collect laboratory historical failure data from a laboratory historical failure database; Preprocess the collected laboratory historical failure data; S322: performing feature extraction; For laboratory environment data, time series analysis method is used to extract features; For laboratory equipment information, the image feature extraction method is used to extract edge features, color features, and texture features from images of equipment failure. A 3D convolutional neural network method is used to extract the spatiotemporal features of equipment failure from the video of equipment failure. The extracted features and the corresponding fault type and fault location are saved as a fault sample data in a data set, wherein the data set contains multiple fault sample data; S323: Model training; The dataset is divided into a training set and a validation set in proportion. The training set is input into the machine learning model. The supervised learning method is used to train the model by taking the fault type and fault location in each fault sample data as labels and the corresponding features as inputs. Furthermore, the features in each fault sample data are associated with the corresponding fault type and fault location; After the training is completed, the validation set is used to evaluate the model and optimize the machine learning model parameters until the machine learning model parameters are trained and the expected effect is achieved, thus obtaining a trained machine learning model. S324: Perform fault diagnosis; The data processing and analysis module inputs the laboratory environment data and laboratory equipment information in the laboratory data set obtained in step S31 into the fault diagnosis model in real time. The machine learning model makes a judgment based on the learned features, fault type, and location. If the machine learning model identifies that there is no fault, the fault diagnosis result output is "no fault"; If the machine learning model identifies a fault, the output fault diagnosis results are the fault type and fault location.
5. The automatic laboratory fault diagnosis method according to claim 1, wherein: The fault report includes the fault time, corresponding laboratory environment data and laboratory equipment data, fault type and location, fault cause analysis, preliminary repair suggestions, and preventive measures. The fault report will be notified to relevant staff via email or text message.
6. A laboratory fault automatic diagnosis system for implementing the laboratory fault automatic diagnosis method, characterized in that: The laboratory fault automatic diagnosis system includes a sensor module, a machine vision module, a data processing and analysis module, and a fault reporting and suggestion module; The sensor module is used to collect laboratory environment data in real time; The machine vision module is used to collect images or videos of equipment in the laboratory in real time; The data processing and analysis module is used to process and analyze sensor data, device images or videos in real time to diagnose the specific fault type and fault location; The fault reporting and suggestion module is used to generate a fault report based on the diagnosed fault type and fault location, and provide preliminary solution suggestions.