Equipment inspection demand prediction method and device, terminal and storage medium

Abstract

The invention discloses an equipment inspection demand prediction method and device, a terminal and a storage medium, and the method comprises the steps: collecting the real-time operation data of to-be-inspected equipment and the environment data of the environment where the to-be-inspected equipment is located through preset monitoring equipment, and obtaining the historical data of the to-be-inspected equipment; performing multi-dimensional data analysis according to the real-time operation data, the environment data and the historical data to obtain an equipment analysis result; and creating a demand prediction model, inputting the equipment analysis result into the demand prediction model to calculate an inspection demand score of the to-be-inspected equipment, performing inspection marking on the to-be-inspected equipment according to the inspection demand score, and generating an inspection route according to the inspection mark. According to the method, the potential fault of the equipment can be predicted more accurately, the unnecessary inspection frequency is reduced, and the equipment management and maintenance efficiency is improved.

Description

Technical Field

[0001] The present application relates to the technical field of equipment inspection demand prediction, and in particular to an equipment inspection demand prediction method, device, terminal and storage medium. Background Art

[0002] With the continuous improvement of industrial automation, the importance of equipment management has become increasingly prominent. Traditional equipment inspection usually relies on manual experience, which is not only time-consuming and labor-intensive, but also difficult to fully cover all possible fault points. In recent years, with the improvement of Internet of Things technology and big data processing capabilities, more and more research has begun to focus on how to use these emerging technologies to improve the equipment inspection process. Most of the existing equipment management systems focus on single-dimensional data monitoring and lack the ability to comprehensively analyze multi-dimensional data, which limits their effectiveness and accuracy in practical applications.

[0003] At present, there are two main methods commonly used in the industry to meet the needs of equipment inspection: one is the rule-based inspection strategy, which sets a series of fixed inspection standards and schedules, and conducts inspections at a predetermined time period regardless of the actual status of the equipment; the other is the status-based inspection strategy, which makes decisions based on the current status of the equipment, but is often limited to certain specific operating parameters and fails to fully consider the impact of equipment historical data and environmental factors. Although the former is simple and easy to implement, it is inflexible and prone to waste resources; although the latter is more accurate, it is still not comprehensive enough in a complex and changing working environment.

[0004] Therefore, existing equipment inspection strategies generally have problems such as poor flexibility, high cost, and low coverage. On the one hand, rule-based inspections check equipment according to a set time period. However, this method is somewhat blind and cannot detect sudden failures in a timely manner. On the other hand, status-based inspections perform poorly under complex equipment operating conditions due to the lack of comprehensive consideration of historical data and environmental factors. In addition, traditional manual inspections rely on personal experience and skills, are greatly affected by subjective factors, and are prone to deviations.

[0005] Therefore, how to improve the equipment inspection process and accurately predict the demand for equipment inspection, thereby improving the accuracy and timeliness of the inspection work, has become an urgent problem to be solved. Summary of the invention

[0006] In order to more accurately predict potential equipment failures, reduce unnecessary inspection times, and improve equipment management and maintenance efficiency, the present application provides a method, device, terminal, and storage medium for predicting equipment inspection needs.

[0007] In the first aspect, the present application provides a method for predicting equipment inspection demand, which adopts the following technical solution: A method for predicting equipment inspection demand comprises the following steps: Collect the real-time operation data of the equipment to be inspected and the environmental data of the environment in which the equipment to be inspected is located through the preset monitoring equipment, and obtain the historical data of the equipment to be inspected; Performing multi-dimensional data analysis based on the real-time operation data, the environmental data, and the historical data to obtain equipment analysis results; Creating a demand forecasting model, inputting the equipment analysis results into the demand forecasting model to calculate the inspection demand score of the equipment to be inspected, The demand forecasting model creation process includes setting an initial demand forecasting model, the initial demand forecasting model is in the form of a linear regression model, and the basic form of the initial demand forecasting model is: ; in, is the target value, , , , is an input feature corresponding to the device analysis result, , , , are the weight coefficients corresponding to the input features, is the bias term; Creating a training data set, training the initial demand forecasting model according to the training data set, calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm, and obtaining the demand forecasting model by combining the trained initial demand forecasting model; The equipment to be inspected is marked for inspection according to the inspection demand score, and an inspection route is generated according to the inspection mark.

[0008] By adopting the above technical solutions, the real-time operation data, environmental data and historical data of the equipment to be inspected can be fully collected, and detailed equipment analysis results can be generated through multi-dimensional data analysis. Based on these analysis results, the demand prediction model is created and trained to accurately calculate the inspection demand score of the equipment to be inspected, accurately predict the potential failure of the equipment, and improve the efficiency of equipment management and maintenance. By marking the inspection demand score and generating the optimal inspection route, the efficiency and accuracy of equipment inspection can be effectively improved, unnecessary inspection workload can be reduced, maintenance costs can be reduced, and the safe and reliable operation of the equipment can be ensured.

[0009] Preferably, the real-time operation data includes current indicator data of various indicators of the equipment to be inspected, and the historical data includes regularly acquired historical indicator data of various indicators of the equipment to be inspected, historical failures and the number of historical failures of the equipment to be inspected, and the total operation time of the equipment to be inspected; The equipment analysis result includes a real-time anomaly score and a load index obtained based on the real-time operation data, a real-time environment score obtained based on the environment data, and a fault frequency and a historical fault severity obtained based on the historical data.

[0010] By adopting the above technical solution, the real-time operation data and historical data of the equipment to be inspected can be fully collected, so as to more accurately evaluate the current status and potential risks of the equipment, effectively improve the demand forecast accuracy of equipment inspection, ensure the efficiency and pertinence of the inspection work, and reduce equipment maintenance costs and downtime risks.

[0011] Preferably, performing multidimensional data analysis based on the real-time operation data, the environmental data and the historical data to obtain the equipment analysis result specifically includes the following steps: Performing data analysis on the real-time operation data by using a Z-score normalization method to obtain a real-time operation score, comparing the real-time operation score with a preset standard operation threshold, and marking the real-time operation score as a real-time abnormal score when the real-time operation score exceeds the standard operation threshold; Performing data analysis on the environmental data by using a Z-score standardization method and obtaining a real-time environmental score; Calculate the historical failure frequency of the equipment to be inspected based on the historical failure times and the total operating time; The current usage time and the maximum rated usage time of the device to be inspected are obtained, and the load index of the device to be inspected is calculated based on the current usage time and the maximum rated usage time.

[0012] By adopting the above technical solution and calculating multi-dimensional data, it is possible to conduct a comprehensive and accurate analysis of the real-time operation data, environmental data and historical data of the equipment, thereby ensuring the comprehensiveness of the inspection demand forecast, improving the accuracy of the inspection demand forecast, and reducing the risk of major accidents caused by failure to timely discover potential problems.

[0013] Preferably, the equipment analysis result also includes the historical fault severity of the equipment to be inspected, and the generation of the historical fault severity specifically includes the following steps: Obtaining the repair time of each of the historical faults in the historical data, scoring the historical faults according to the repair time and in combination with a preset fault scoring standard to obtain a historical fault impact score, and using a weighted average method to calculate the severity of the historical faults of the equipment to be inspected; Right now ; in, is the severity of historical faults, Assign a score to the historical fault, is the number of historical failures.

[0014] By adopting the above technical solution, the severity of historical faults is set as an influencing factor, and the severity of historical faults of the inspection equipment is evaluated. This not only takes into account the impact of a single fault, but also fully reflects the impact of multiple faults on the overall reliability of the equipment, thereby improving the accuracy of inspection demand prediction and making the final prediction result more comprehensive.

[0015] Preferably, the creating of a training data set, training the initial demand forecasting model according to the training data set, and calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm specifically include the following steps: Creating a training data set for the initial demand forecasting model, the training data set includes the input features and target values, the input features include the real-time anomaly score, the historical fault frequency, the historical fault severity, the real-time environment score, and the load index, and the target value is the inspection demand score; Initializing the weight coefficient and the bias term, and calculating the error between the initial demand forecast model and the target value in combination with a preset loss function, The calculation formula of the error is: ; ; in, is the number of samples in the training dataset; The model is The predicted value of samples; It is The true target value of samples, It is The first input feature samples.

[0016] By adopting the above technical solution, the training data set is accurately constructed to ensure the effective matching of input features and target values; by initializing the weight coefficients and bias terms, and combining the preset loss function to calculate the error between the initial demand forecast model and the true target value, the performance of the model can be effectively evaluated. This method improves the accuracy and reliability of the model, thereby improving the accuracy of equipment inspection demand forecasting.

[0017] Preferably, the method further comprises the following steps: Calculating the partial derivative of the loss function with respect to the weight coefficient and the bias term, The partial derivative is calculated as follows: ; ; Indicates The first input feature Sample values, Represents the loss function with respect to The weight coefficient of the input feature The partial derivative of represents the partial derivative of the loss function with respect to the bias term; According to a preset gradient descent algorithm, the weight coefficient and the bias term are optimized iteratively in combination with the partial derivative of the weight coefficient and the partial derivative of the bias term until a set of the weight coefficient and the bias term that minimize the error is obtained.

[0018] By adopting the above technical solution and using the gradient descent algorithm to effectively optimize the loss function for weight coefficients and bias terms, not only the accuracy of the demand forecasting model is improved, but also it is ensured that the model can better adapt to different inspection scenarios in practical applications, thereby improving the demand forecasting accuracy and efficiency of equipment inspection.

[0019] Preferably, marking the equipment to be inspected according to the inspection demand score, and generating an inspection route according to the inspection mark, specifically comprises the following steps: Comparing the inspection demand score with a preset inspection threshold; When the inspection requirement score is less than the inspection threshold, the device to be inspected is marked as not requiring inspection; when the inspection requirement score is not less than the inspection threshold, the device to be inspected is marked as requiring inspection; The geographical location information of all the devices to be inspected that carry the marks that need to be inspected is obtained and summarized, and a shortest inspection route is calculated and output based on a preset path optimization algorithm.

[0020] By adopting the above technical solution, whether the equipment to be inspected needs to be inspected is automatically determined based on the calculated inspection demand score, thereby avoiding the subjectivity and errors of human judgment, improving the consistency and reliability of decision-making, and ensuring that high-risk equipment receives timely attention and maintenance; at the same time, through the optimal inspection route planning, the efficiency and scientific nature of the inspection work are further improved.

[0021] In the second aspect, the present application provides a device for predicting equipment inspection demand, which adopts the following technical solution: A device for predicting equipment inspection demand, comprising the following modules: A data acquisition module is used to collect real-time operation data of the equipment to be inspected and environmental data of the environment in which the equipment to be inspected is located through a preset monitoring device, and to obtain historical data of the equipment to be inspected; A multidimensional data analysis module, used to perform multidimensional data analysis based on the real-time operation data, the environmental data and the historical data to obtain equipment analysis results; An inspection demand score calculation module is used to create a demand prediction model, and calculate the inspection demand score of the equipment to be inspected based on the demand prediction model and the equipment analysis results. The demand forecasting model creation process includes setting an initial demand forecasting model, the initial demand forecasting model is in the form of a linear regression model, and the basic form of the initial demand forecasting model is: ; in, is the target value, , , , is an input feature corresponding to the device analysis result, , , , are the weight coefficients corresponding to the input features, is the bias term; Creating a training data set, training the initial demand forecasting model according to the training data set, calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm, and obtaining the demand forecasting model by combining the trained initial demand forecasting model; The inspection route generation module is used to mark the equipment to be inspected according to the inspection demand score and generate an inspection route according to the inspection mark.

[0022] By adopting the above-mentioned technical means, a set of intelligent assessment system for equipment inspection needs based on multi-dimensional data analysis and artificial intelligence technology has been built, which provides the necessary software technical support for the inspection demand prediction of factory equipment and ensures the safety and reliability of the site.

[0023] In a third aspect, the present application provides a smart terminal, which adopts the following technical solution: An intelligent terminal includes a memory and a processor, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the equipment inspection demand prediction method as described above.

[0024] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium, wherein at least one instruction, at least one program, code set or instruction set is stored in the computer-readable storage medium, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the equipment inspection demand prediction method as described above.

[0025] In summary, this application has at least the following beneficial effects: (1) This application uses multi-dimensional data analysis and demand forecasting models to intelligently evaluate equipment inspection needs, comprehensively consider the real-time operating data, environmental data and historical data of the equipment, and more accurately predict potential equipment failures, thereby improving the accuracy and reliability of inspection demand forecasts and improving equipment management and maintenance efficiency.

[0026] (2) This application uses a multi-factor weighted linear regression model to ensure that the scoring process is flexible and scalable, while scientifically evaluating the necessity of equipment inspections, reducing unnecessary inspection work, and saving manpower and material resources.

[0027] (3) The inspection mark of this application generates the optimal inspection route, ensuring the efficiency and pertinence of the inspection work and improving the overall inspection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a flow chart of a method for predicting equipment inspection demand according to an embodiment of the present application; Figure 2 It is a multi-dimensional data analysis flow chart of the equipment inspection demand prediction method of the embodiment of the present application; Figure 3 is a flowchart of step S3 of the equipment inspection demand prediction method of an embodiment of the present application; Figure 4It is a flowchart of the initial demand prediction model training of the equipment inspection demand prediction method in the embodiment of the present application; Figure 5 It is an architectural diagram of the equipment inspection demand prediction device of an embodiment of the present application. DETAILED DESCRIPTION

[0029] The present application provides a method, device, terminal and storage medium for predicting equipment inspection demand. In order to make the purpose, technical solution and advantages of the present application clearer, the implementation method of the present application will be further described in detail below.

[0030] The following is a further detailed description of an embodiment of a method for predicting equipment inspection demand of the present application in conjunction with the drawings in the specification.

[0031] A method for predicting equipment inspection demand in this application, such as Figure 1 As shown, the following steps are included: S1. Collect the real-time operation data of the equipment to be inspected and the environmental data of the environment in which the equipment to be inspected is located through the preset monitoring equipment, and obtain the historical data of the equipment to be inspected.

[0032] In this embodiment, various sensors and IoT devices are used to monitor the real-time operating status of the equipment to be inspected, and real-time operating data and environmental data of each device are collected regularly, and historical data and maintenance records of the equipment are collected from the equipment maintenance information system.

[0033] Real-time operation data includes current indicator data of various indicators of the equipment to be inspected, including temperature, vibration, speed, current, pressure and current operating time; environmental data includes temperature, humidity and dust information of the environment in which the equipment is located; historical data includes regularly acquired historical indicator data of various indicators of the equipment to be inspected, historical failures of the equipment to be inspected and the number of historical failures, and the total operating time of the equipment to be inspected; maintenance records include the number of maintenance times and parts replacement records.

[0034] S2. Perform multidimensional data analysis based on real-time operation data, environmental data, and historical data to obtain equipment analysis results. The equipment analysis results include real-time anomaly scores and load indexes obtained based on real-time operation data, real-time environmental scores obtained based on environmental data, and fault frequencies and historical fault severity obtained based on historical data. Multidimensional data analysis is as follows: Figure 2 As shown, the specific steps include: S21. Perform data analysis on the real-time operation data through the Z-score normalization method to obtain a real-time operation score, compare the real-time operation score with a preset standard operation threshold, and when the real-time operation score exceeds the standard operation threshold, mark the real-time operation score as a real-time abnormal score; when the real-time operation score does not exceed the standard operation threshold, do not operate the real-time operation score, and the real-time operation score is not subsequently calculated.

[0035] In a specific implementation method, taking the Z-score standardization calculation of the temperature index as an example, ; in, It is the real-time running score of the temperature index of the current device to be tested. is the current temperature reading, and They are the historical average and standard deviation of the temperature of the equipment to be inspected. and Calculated based on historical data.

[0036] if If the set temperature exceeds the standard threshold, it is considered that the temperature of the device is abnormal, affecting the inspection needs. Marked as real-time anomaly score.

[0037] S22. Analyze environmental data using the Z-score standardization method and obtain a real-time environmental score.

[0038] Data analysis of environmental data includes monitoring the temperature, humidity, dust, and noise of the environment in which the equipment to be tested is located. The real-time environmental score is obtained by using the Z-score method, which will not be repeated here.

[0039] The worse the environmental conditions, the faster the equipment wears out, and the inspection needs score increases accordingly.

[0040] S23. Calculate the historical failure frequency of the equipment to be inspected based on the number of historical failures and the total operating time. ,Right now, .

[0041] S24. Obtain the current usage time and the maximum rated usage time of the equipment to be inspected, and calculate the load index of the equipment to be inspected based on the current usage time and the maximum rated usage time.

[0042] If the equipment to be inspected runs at high load for a long time, the inspection demand will increase. Use the following formula to calculate the load index: ; in, For the load index, the current usage time is obtained from the real-time operation data, and the maximum rated usage time is directly called from the storage of the equipment.

[0043] The higher the load index, the higher the inspection demand score.

[0044] S25. The equipment analysis results also include the historical fault severity of the equipment to be inspected. The historical fault severity reflects the impact of each fault on the equipment operation.

[0045] The generation of historical fault severity specifically includes the following steps: S251. Obtain the repair time of each historical failure in the historical data.

[0046] S252. Calculate the average repair time of faults in historical records based on the repair time, and score the historical faults in combination with a preset fault scoring standard to obtain a historical fault impact score.

[0047] In this embodiment, the repair time is scored according to the standards of mild (1 point), moderate (2 points), and severe (3 points). A threshold of the repair time is set for each severity level. When the repair time falls within the threshold of a certain level, the corresponding historical fault is scored to obtain a historical fault impact score.

[0048] In another specific implementable method, a fault scoring standard can be formulated based on multiple dimensions of the fault, including the scope of impact, duration, and difficulty of repair. Each dimension has a full score of 3 points, and is scored according to the standards of mild (1 point), moderate (2 points), and severe (3 points).

[0049] For example, a mild impact is a local impact, where only a single device or function is affected; a moderate impact is a system impact, where multiple devices or functions are affected; and a severe impact is a comprehensive impact, where the entire production line or system is paralyzed.

[0050] The duration of mild is short: recovery within 1 hour; moderate is moderate: recovery within a day; severe is long: more than a day or lasting for several days.

[0051] The repair difficulty ranges from mild to simple: it can be solved immediately without replacing parts; moderate to medium: it requires professional intervention, but there is no need to replace key parts; severe to complex: it requires long-term diagnosis, replacement of key parts or system reconstruction. The above data can be determined based on maintenance records.

[0052] Security risks are also included as an additional item, with a maximum score of 1 point: 0 points when there is no safety risk; When there is a safety risk, such as possible loss of life or environmental pollution: 1 point.

[0053] The scores for each dimension are added together to obtain the impact score for each historical fault.

[0054] S253, using a weighted average method to calculate the severity of historical faults of the equipment to be inspected; Right now ; in, is the severity of historical faults, Assign a score to the historical fault, is the number of historical failures.

[0055] S3. Create a demand forecasting model and input the equipment analysis results into the demand forecasting model to calculate the inspection demand score of the equipment to be inspected, such as Figure 3 As shown, the specific steps include: S31. Setting an initial demand forecasting model. In this embodiment, the initial demand forecasting model is set as a weighted linear regression model. The weighted linear regression model is the prototype of the demand forecasting model to be created.

[0056] The basic form of the weighted linear regression model is, ; Score the inspection needs. , , , For each input feature, , , , are the weight coefficients that need to be determined for each input feature. is the bias term.

[0057] S32, calculating the weight coefficient and bias term of each main feature according to the preset optimization iterative algorithm, the calculation method of the weight coefficient and bias term specifically includes the following steps: Create a training dataset for a weighted linear regression model, such as Figure 4 As shown, the initial demand prediction model is trained based on the training data set. The training data set includes input features and target values. The input features include real-time anomaly score, historical fault frequency, historical fault severity, real-time environment score and load index. The data in the training data set is obtained based on the past historical data of all equipment, and the target value is the inspection demand score.

[0058] S33, initializing the weight coefficient and the bias term, calculating the error between the initial demand forecast model and the target value in combination with a preset loss function, optimizing the weight coefficient and the bias term, and finally obtaining the optimal weight coefficient and bias term. Specifically, the steps include: S331, weight coefficient , , , and the bias term Initialize to a small random value, 0.1 in this embodiment.

[0059] S332. Setting a loss function to measure the error between the model prediction value and the actual target value. The loss function in this embodiment is the mean square error (MSE).

[0060] The calculation formula of mean square error is: ; ; in, is the number of samples in the training dataset; The model is The predicted value of samples; It is The true target value of samples, It is The first input feature samples.

[0061] S333. Minimize the loss function and find the optimal weight coefficient , , , and the bias term , so that the error between the predicted value and the actual value of the model is as small as possible. In this embodiment, the gradient descent algorithm is selected to optimize the weight and bias terms.

[0062] The formula for the gradient descent algorithm is: ; ; in, is the learning rate, which is used to control the size of each adjustment step, is the loss function for the weight coefficient The partial derivative of , which indicates the impact of changes in weight coefficients on the loss; is the loss function with respect to the bias term The partial derivative of .

[0063] This embodiment dynamically adjusts the learning rate according to the size and direction of the gradient. Specifically, the Adam optimizer is used to adaptively adjust the learning rate according to historical gradient information.

[0064] S334. Further, the above loss function has a weight coefficient The partial derivative of the loss function with respect to the bias term The update calculation process of the partial derivative of is as follows: ; ; in, Indicates The first input feature Sample values.

[0065] S335, combining the partial derivative of the weight coefficient and the partial derivative of the bias term to optimize the weight coefficient and the bias term iteratively, through multiple iterative training, the weight coefficient , , , and the bias term Adjust to a value that minimizes the error of the loss function.

[0066] These weight coefficients and bias terms represent the relative impact of each input feature on the inspection demand score.

[0067] In order to make the above description clearer, a specific embodiment is used for description below: Assume that the following five samples are used for model training (for simplicity, only four features are used here): <![CDATA[Real-time anomaly score X1]]> <![CDATA[Historical failure frequency X2]]> <![CDATA[Historical fault severity X3]]> <![CDATA[Real-time environmental score X4]]> <![CDATA[True inspection score Y true > 1.5 0.05 2.0 0.8 2.5 0.8 0.02 1.5 0.5 5.0 2.0 0.08 2.5 0.9 8.0 0.7 0.03 1.2 0.4 4.5 1.2 0.06 2.1 0.7 6.5

[0068] Initial weight =0.1, =0.1, =0.1, = 0.1, bias term =0.1.

[0069] After multiple rounds of gradient descent updates, the weights are adjusted to: =0.6, =0.4, =0.5, = 0.2, bias term =0.3.

[0070] Prediction of inspection demand score for the first sample: =0.6×1.5+0.4×0.05+0.5×2.0+0.2×0.8+0.3=2.38.

[0071] The actual inspection score is 2.5, which is relatively close.

[0072] S34. A demand forecasting model is obtained by combining the trained initial demand forecasting model, that is, the demand forecasting model is obtained according to the calculated weight coefficients and bias items. The demand forecasting model is a weighted linear regression model.

[0073] The real-time anomaly score, historical fault frequency, historical fault severity, real-time environmental score, and load index are input into the weighted linear regression model as the main features and the inspection demand score is calculated.

[0074] The calculation formula for the inspection demand score in this embodiment is: ; in, Score the inspection needs. is the real-time anomaly score, is the weight coefficient of the real-time anomaly score; is the historical fault frequency, is the weight coefficient of historical fault frequency; is the severity of historical faults, is the weight coefficient of the severity of historical faults; Score for real-time environment, is the weight coefficient of the real-time environment score; is the load index, is the weight coefficient of the load index.

[0075] It is a bias item used to adjust the baseline of the inspection demand score.

[0076] S35. Input the equipment analysis results into the demand prediction model to calculate the inspection demand score of the equipment to be inspected.

[0077] S4. Mark the equipment to be inspected according to the inspection demand score, and generate an inspection route according to the inspection mark, which specifically includes the following steps: Compare the inspection need score with the preset inspection threshold; When the inspection demand score is less than the inspection threshold, the device to be inspected is marked as not requiring inspection; when the inspection demand score is not less than the inspection threshold, the device to be inspected is marked as requiring inspection; The geographic location information of all devices to be inspected that carry the marks that need to be inspected is obtained and summarized, and the shortest inspection route is calculated and output based on the preset path optimization algorithm.

[0078] In a specific implementation, the steps of calculating the shortest inspection route are: After obtaining all the devices to be inspected in a certain area, the devices to be inspected and their inspection demand scores are summarized, and all the devices to be inspected are divided into different levels according to the inspection demand scores, namely high risk level, medium risk level and low risk level.

[0079] Determine the priority of all devices that need to be inspected based on their levels.

[0080] Based on the historical failure frequency of each device, set the inspection frequency for each device, and make dynamic adjustments based on the inspection demand score.

[0081] Obtain the geographic location information of each device, which is the latitude and longitude coordinates or relative position relationship; Using the Dijkstra algorithm, the geographical location information, inspection priority, inspection frequency, etc. of the equipment are input as input parameters into the Dijkstra algorithm.

[0082] The Dijkstra algorithm calculates and generates the shortest inspection path from the starting point to all devices that need to be inspected and then returns to the starting point based on the input parameters; if the inspector needs to start from multiple starting points, the shortest path from each starting point to the device he is responsible for inspecting is calculated separately and merged into a complete inspection path.

[0083] Based on the generated shortest inspection path, a detailed inspection plan is generated, which includes inspection time, inspection personnel, inspection equipment, and inspection path.

[0084] Based on the same inventive concept as above, the present application also discloses a device for predicting equipment inspection demand, the structure of which is as follows: Figure 5 As shown, the device includes the following modules: The data acquisition module is used to collect the real-time operation data of the equipment to be inspected and the environmental data of the environment in which the equipment to be inspected is located through the preset monitoring equipment, and obtain the historical data of the equipment to be inspected; Multidimensional data analysis module, used to perform multidimensional data analysis based on real-time operation data, environmental data and historical data to obtain equipment analysis results; The inspection demand score calculation module is used to create a demand forecasting model and input the equipment analysis results into the demand forecasting model to calculate the inspection demand score of the equipment to be inspected. The process of creating a demand forecasting model includes setting an initial demand forecasting model. The form of the initial demand forecasting model is a linear regression model. The basic form of the initial demand forecasting model is: ; in, is the target value, , , , is an input feature corresponding to the device analysis result, , , , are the weight coefficients corresponding to the input features, is the bias term; Create a training data set, train the initial demand forecasting model based on the training data set, calculate the weight coefficient and bias term based on the preset optimization iterative algorithm, and obtain the demand forecasting model by combining the trained initial demand forecasting model; The inspection route generation module is used to mark the equipment to be inspected according to the inspection demand score and generate the inspection route according to the inspection mark.

[0085] In a specific implementation scheme, the multidimensional data analysis module includes the following units: A first multidimensional data analysis unit is used to perform data analysis on the real-time operation data by using a Z-score normalization method to obtain a real-time operation score, compare the real-time operation score with a preset standard operation threshold, and mark the real-time operation score as a real-time abnormal score when the real-time operation score exceeds the standard operation threshold; The second multidimensional data analysis unit is used to analyze the environmental data by using the Z-score standardization method and obtain a real-time environmental score; The third multi-dimensional data analysis unit is used to calculate the historical failure frequency of the equipment to be inspected based on the number of historical failures and the total operating time; The fourth multi-dimensional data analysis unit is used to obtain the current usage time and the maximum rated usage time of the equipment to be inspected, and calculate the load index of the equipment to be inspected based on the current usage time and the maximum rated usage time.

[0086] The fifth multidimensional data analysis unit is used to obtain the repair time of each historical fault in the historical data, score the historical fault according to the repair time and the preset fault scoring standard, obtain the historical fault impact score, and use the weighted average method to calculate the severity of the historical fault of the equipment to be inspected; Right now ; in, is the severity of historical faults, Score the historical fault as The number of historical failures.

[0087] In a specific implementation scheme, the inspection demand score calculation module includes the following units: A first inspection demand score calculation unit is used to create a training data set for an initial demand prediction model, the training data set includes input features and target values, the input features include real-time anomaly score, historical fault frequency, historical fault severity, real-time environment score and load index, and the target value is the inspection demand score; The second inspection demand score calculation unit is used to initialize the weight coefficient and bias term, and calculate the error between the initial demand forecast model and the target value in combination with the preset loss function. The error calculation formula is: ; ; in, is the number of samples in the training dataset; The model is The predicted value of samples; It is The true target value of samples, It is The first input feature samples.

[0088] The third inspection demand score calculation unit is used to calculate the partial derivative of the loss function with respect to the weight coefficient and the bias term. The partial derivative is calculated as, ; ; Indicates The first input feature Sample values, Represents the loss function with respect to The weight coefficient of the input features The partial derivative of Represents the partial derivative of the loss function with respect to the bias term; The fourth inspection demand score calculation unit is used to optimize and iterate the weight coefficient and the bias term according to a preset gradient descent algorithm, combined with the partial derivative of the weight coefficient and the partial derivative of the bias term, until a set of weight coefficients and bias terms that minimize the error is obtained.

[0089] In a specific implementation scheme, the inspection route generation module includes the following units: A first inspection route generating unit, used for comparing the inspection demand score with a preset inspection threshold; A second inspection route generating unit is used to mark the device to be inspected as not requiring inspection when the inspection demand score is less than the inspection threshold, and to mark the device to be inspected as requiring inspection when the inspection demand score is not less than the inspection threshold; The third inspection route generating unit is used to obtain and summarize the geographical location information of all the devices to be inspected that carry the marks that need to be inspected, calculate a shortest inspection route according to a preset path optimization algorithm, and output it.

[0090] From the above functional introduction, it can be seen that the present application can comprehensively collect the real-time operation data, environmental data and historical data of the equipment to be inspected, and use multi-dimensional data analysis to obtain detailed equipment analysis results; further, by creating and training a demand prediction model, the inspection demand score of the equipment to be inspected can be accurately calculated, thereby effectively identifying the equipment that needs to be inspected; finally, inspection marks are generated based on the inspection demand score, and the shortest inspection route is generated in combination with geographic information optimization, which improves inspection efficiency and accuracy and reduces inspection costs.

[0091] Based on the same inventive concept mentioned above, an embodiment of the present application also discloses a computer-readable storage medium, which stores at least one instruction, at least one program, code set or instruction set. The at least one instruction, at least one program, code set or instruction set can be loaded and executed by a processor to implement the equipment inspection demand prediction method provided in the above method embodiment.

[0092] Also based on the same inventive concept mentioned above, an embodiment of the present application also discloses a computer-readable storage medium, which stores at least one instruction, at least one program, code set or instruction set. The at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the equipment inspection demand prediction method as described above.

[0093] Those skilled in the art will appreciate that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in the computer-readable storage medium, which may include, for example, various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0094] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting equipment inspection demand, characterized in that: The steps include: Collect the real-time operation data of the equipment to be inspected and the environmental data of the environment in which the equipment to be inspected is located through the preset monitoring equipment, and obtain the historical data of the equipment to be inspected; Performing multi-dimensional data analysis based on the real-time operation data, the environmental data, and the historical data to obtain equipment analysis results; Creating a demand forecasting model, inputting the equipment analysis results into the demand forecasting model to calculate the inspection demand score of the equipment to be inspected, The demand forecasting model creation process includes setting an initial demand forecasting model, the initial demand forecasting model is in the form of a linear regression model, and the basic form of the initial demand forecasting model is: ; in, is the target value, , , , is an input feature corresponding to the device analysis result, , , , are the weight coefficients corresponding to the input features, is the bias term; Creating a training data set, training the initial demand forecasting model according to the training data set, calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm, and obtaining the demand forecasting model by combining the trained initial demand forecasting model; The equipment to be inspected is marked for inspection according to the inspection demand score, and an inspection route is generated according to the inspection mark.

2. The equipment inspection demand prediction method according to claim 1 is characterized in that: The real-time operation data includes current indicator data of various indicators of the equipment to be inspected, and the historical data includes regularly acquired historical indicator data of various indicators of the equipment to be inspected, historical faults and the number of historical faults of the equipment to be inspected, and the total operation time of the equipment to be inspected; The equipment analysis result includes a real-time anomaly score and a load index obtained based on the real-time operation data, a real-time environment score obtained based on the environment data, and a fault frequency and a historical fault severity obtained based on the historical data.

3. The equipment inspection demand prediction method according to claim 2 is characterized in that: The multi-dimensional data analysis is performed based on the real-time operation data, the environmental data and the historical data to obtain the equipment analysis result, which specifically includes the following steps: Performing data analysis on the real-time operation data by using a Z-score normalization method to obtain a real-time operation score, comparing the real-time operation score with a preset standard operation threshold, and marking the real-time operation score as a real-time abnormal score when the real-time operation score exceeds the standard operation threshold; Performing data analysis on the environmental data by using a Z-score standardization method and obtaining a real-time environmental score; Calculate the historical failure frequency of the equipment to be inspected based on the historical failure times and the total operating time; The current usage time and the maximum rated usage time of the device to be inspected are obtained, and the load index of the device to be inspected is calculated based on the current usage time and the maximum rated usage time.

4. The equipment inspection demand prediction method according to claim 3 is characterized in that: The equipment analysis result also includes the historical fault severity of the equipment to be inspected, and the generation of the historical fault severity specifically includes the following steps: Obtaining the repair time of each of the historical faults in the historical data, scoring the historical faults according to the repair time and in combination with a preset fault scoring standard to obtain a historical fault impact score, and using a weighted average method to calculate the severity of the historical faults of the equipment to be inspected; Right now ; in, is the severity of historical faults, Assign a score to the historical fault, is the number of historical failures.

5. The equipment inspection demand prediction method according to claim 2, characterized in that: The step of creating a training data set, training the initial demand forecasting model according to the training data set, and calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm specifically includes the following steps: Creating a training data set for the initial demand forecasting model, the training data set includes the input features and target values, the input features include the real-time anomaly score, the historical fault frequency, the historical fault severity, the real-time environment score, and the load index, and the target value is the inspection demand score; Initializing the weight coefficient and the bias term, and calculating the error between the initial demand forecast model and the target value in combination with a preset loss function, The calculation formula of the error is: ； ； in, is the number of samples in the training dataset; The model is The predicted value of samples; It is The true target value of samples, It is The first input feature samples.

6. The equipment inspection demand prediction method according to claim 5, characterized in that: The following steps are also included: Calculating the partial derivative of the loss function with respect to the weight coefficient and the bias term, The partial derivative is calculated as follows: ； ； Indicates The first input feature Sample values, Represents the loss function with respect to The weight coefficient of the input feature The partial derivative of represents the partial derivative of the loss function with respect to the bias term; According to a preset gradient descent algorithm, the weight coefficient and the bias term are optimized iteratively in combination with the partial derivative of the weight coefficient and the partial derivative of the bias term until a set of the weight coefficient and the bias term that minimize the error is obtained.

7. The equipment inspection demand prediction method according to claim 1, characterized in that: The step of marking the equipment to be inspected according to the inspection demand score and generating an inspection route according to the inspection mark specifically includes the following steps: Comparing the inspection demand score with a preset inspection threshold; When the inspection requirement score is less than the inspection threshold, the device to be inspected is marked as not requiring inspection; when the inspection requirement score is not less than the inspection threshold, the device to be inspected is marked as requiring inspection; The geographical location information of all the devices to be inspected that carry the marks that need to be inspected is obtained and summarized, and a shortest inspection route is calculated and output based on a preset path optimization algorithm.

8. A device for predicting equipment inspection demand, characterized in that: Includes the following modules: A data acquisition module is used to collect real-time operation data of the equipment to be inspected and environmental data of the environment in which the equipment to be inspected is located through a preset monitoring device, and to obtain historical data of the equipment to be inspected; A multidimensional data analysis module, used to perform multidimensional data analysis based on the real-time operation data, the environmental data and the historical data to obtain equipment analysis results; An inspection demand score calculation module is used to create a demand prediction model, and calculate the inspection demand score of the equipment to be inspected based on the demand prediction model and the equipment analysis results. The demand forecasting model creation process includes setting an initial demand forecasting model, the initial demand forecasting model is in the form of a linear regression model, and the basic form of the initial demand forecasting model is: ； in, is the target value, , , , is an input feature corresponding to the device analysis result, , , , are the weight coefficients corresponding to the input features, is the bias term; Creating a training data set, training the initial demand forecasting model according to the training data set, calculating the weight coefficient and the bias term according to a preset optimization iterative algorithm, and obtaining the demand forecasting model by combining the trained initial demand forecasting model; The inspection route generation module is used to mark the equipment to be inspected according to the inspection demand score and generate an inspection route according to the inspection mark.

9. An intelligent terminal, characterized in that: It includes a memory and a processor, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the equipment inspection demand prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The readable storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the equipment inspection demand prediction method as described in any one of claims 1 to 7.

Images