Equipment early warning system based on port digitization

Abstract

The invention belongs to the technical field of equipment early warning, and discloses an equipment early warning system based on port digitization, and the system comprises a data processing module which is used for carrying out the quantitative processing of an operation task into an operation quantitative index, and extracting the operation features of equipment; the prediction model module is used for constructing an equipment operation prediction model according to the historical data; the real-time prediction module is used for acquiring a quantitative index of a next operation task and predicting equipment operation characteristics; the prediction correction module is used for correcting the prediction result according to the historical deviation; the task evaluation module is used for evaluating task adaptability from three dimensions of safety, maintenance requirements and timeliness; and the alarm module is used for performing graded early warning according to the task adaptability. Through quantitative analysis of job task characteristics and consideration of equipment individual differences, accurate prediction of future job risks is realized, and through a multi-dimensional evaluation and grading early warning mechanism, the efficiency and accuracy of risk disposal are significantly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment warning, and more specifically, the present invention relates to an equipment warning system based on port digitization. Background Art

[0002] As a core equipment for port loading and unloading operations, the safe and stable operation of container cranes is crucial for ensuring port production efficiency and the smoothness of the international logistics chain. Container cranes usually bear high-intensity and high-frequency operation loads, and their working conditions directly affect the throughput capacity and service quality of ports. Therefore, the condition monitoring and risk warning of container cranes have become an important link in port equipment management.

[0003] At present, the warning systems commonly used in container crane management mainly conduct risk prediction based on the historical operation data and current state parameters of the equipment, such as monitoring key indicators such as motor temperature and hydraulic system pressure, and obtaining the change trend of a certain indicator from the historical operation data. When it exceeds the preset threshold, a corresponding alarm mechanism is triggered. Although this current-state-based warning method can respond to the immediate abnormal conditions of the equipment, it has certain limitations. First of all, the traditional warning system does not consider the characteristics of the operation tasks to be executed by the equipment and cannot predict the possible impact of future operations on the equipment. For example, when the equipment is about to execute a high-load and high-intensity operation task, even if the current equipment state parameters do not exceed the threshold, it may cause equipment failures or safety accidents due to cumulative loads during the operation. In addition, the existing warning systems are relatively single in the dimension of risk assessment, mainly focusing on the safety indicators of the equipment, and insufficiently considering aspects such as maintenance requirements and operation efficiency. This one-sided risk assessment results in poor pertinence of the warning information, making it impossible to conduct refined classification according to the risk type and severity, resulting in over-alarm or under-alarm situations, which affects the efficiency and accuracy of risk disposal. Summary of the Invention

[0004] In order to overcome the above problems of the prior art, the present invention proposes an equipment warning system based on port digitization to solve the above problems.

[0005] The present invention provides the following technical solutions: An equipment warning system based on port digitization, comprising: A data processing module, configured to quantitatively process an operation task into an operation quantization index, and also configured to collect operation data when executing the operation task and extract equipment operation characteristics; A prediction model module, configured to use the historical operation quantization indexes and the corresponding equipment operation characteristics to form a training data set, and construct an equipment operation prediction model according to the training data set, where the equipment operation prediction model is configured to predict equipment operation characteristics according to the operation quantization index; A real-time prediction module, configured to obtain job quantization metrics corresponding to the next job task to be executed, and predict device job characteristics through a device job prediction model; A prediction correction module, configured to obtain a performance deviation vector according to the device job characteristics predicted by the device job prediction model and the corresponding actually extracted device job characteristics in the history of the execution device; correct the predicted device job characteristics according to the performance deviation vector to obtain corrected job characteristic data; A task evaluation module, configured to evaluate the task adaptability of the execution device to the next job task to be executed according to the corrected job characteristic data and the job quantization metrics; An alarm module, configured to perform corresponding early warnings according to the evaluated task adaptability.

[0006] Preferably, the quantization of the job task into job quantization metrics includes: Extracting container data metrics, job time metrics, and job space metrics from the job task; The container data metrics include the total number of containers, the total weight of containers, the average weight of containers, the maximum and minimum weights of containers, and the container weight distribution index, and the container weight distribution index is calculated by the sum of the products of the proportion of the number of containers in each weight interval to the total number and the median of the corresponding interval; the job time metrics include the planned job time; The job space metrics include the maximum and minimum loading and unloading heights, the average loading and unloading height, the loading and unloading height distribution index, the maximum and minimum loading and unloading radii, the average loading and unloading radius, and the loading and unloading radius distribution index, wherein the loading and unloading height distribution index and the loading and unloading radius distribution index are respectively calculated by the sum of the products of the proportion in each height interval and each radius interval and the median of the corresponding interval; Performing standardization processing on the container data metrics, job time metrics, and job space metrics to form job quantization metrics.

[0007] Preferably, the collection of job data during the execution of the job task and the extraction of device job characteristics include: Collecting the main motor temperature, hydraulic system pressure value, main structure stress value, and device vibration frequency during the execution of the job task, and extracting the maximum value among the above values to form device operation characteristics; Collecting the hydraulic oil pollution degree and the wear degree of each connection part during the execution of the job task, and obtaining the corresponding pollution change degree and wear change degree according to the values at the start and completion of the job task to form device loss characteristics; Obtaining the total elapsed time for completing the job task as the task time characteristic; Using the device operation characteristics, device loss characteristics, and task time characteristics to form device job characteristics.

[0008] Preferably, constructing the equipment operation prediction model according to the training data set includes: Performing data cleaning on the historical operation quantization indexes and the corresponding equipment operation characteristics, removing outliers and missing values, and forming an effective training data set; Dividing the effective training data set into a training set and a validation set according to a preset ratio; Constructing a deep neural network model structure, including an input layer, multiple hidden layers, and an output layer, where the number of nodes in the input layer is the same as the dimension of the operation quantization index, and the number of nodes in the output layer is the same as the dimension of the equipment operation characteristics; Using the operation quantization indexes in the training set as inputs and the equipment operation characteristics as output labels, and optimizing the model parameters by using the gradient descent optimization algorithm; Through multiple iterative trainings of the training set data, enabling the model to learn the mapping relationship between the operation quantization indexes and the equipment operation characteristics; During the training process, using the validation set to evaluate the model performance, and terminating the training process when the error value on the validation set no longer decreases for multiple consecutive iterations; Using the cross-validation method to evaluate the model performance and calculating the error index between the predicted value and the actual value; Adjusting the model structure and hyperparameters according to the evaluation results, and repeatedly using the training set for training until the model performance meets the predetermined threshold requirements to obtain the equipment operation prediction model.

[0009] Preferably, the steps for obtaining the performance deviation vector include: Collecting the predicted equipment operation characteristics and the actually extracted equipment operation characteristic data in at least the last five operations of the execution equipment; For the predicted characteristics and the actual characteristics of each operation, calculating the difference between the actual value and the predicted value for each dimension; Arranging the differences of each dimension in chronological order to form a deviation sequence for each dimension; Predicting the expected deviation value of each current dimension according to the deviation sequences of each dimension; Combining the expected deviation values of each dimension to form a performance deviation vector.

[0010] Preferably, predicting the expected deviation value of each current dimension according to the deviation sequences of each dimension includes: Performing outlier detection and removal on the deviation sequences of each dimension; For the deviation sequences of each dimension after removing outliers, calculating the weighted mean, the deviation change rate, and the fluctuation index; Taking the most recent deviation value in the deviation sequence as a benchmark, and obtaining a preliminary deviation in combination with the change rate and the time interval; Judging the fluctuation index, and if the fluctuation index is lower than the set threshold, directly using the preliminary deviation as the expected deviation value; Otherwise, fuse the weighted mean and the preliminary deviation, and take their arithmetic mean as the expected deviation value.

[0011] Preferably, the task adaptability of the evaluation execution device to the next job task to be executed includes: Compare the values of each dimension of the device operation characteristics in the corrected job characteristic data with the corresponding safe operation threshold respectively; mark the dimension exceeding the corresponding safety threshold as the safety risk dimension; Obtain the current hydraulic oil pollution degree and the wear degree data of each connection part of the execution device, and add them to the corresponding pollution change degree and wear change degree in the corrected job characteristic data respectively to obtain the predicted pollution degree and the predicted wear degree; Compare the predicted pollution degree and the predicted wear degree with the corresponding maintenance thresholds respectively, and mark the dimension exceeding the corresponding maintenance threshold as the maintenance risk dimension; Compare the task time characteristic in the corrected job characteristic data with the job time index. If the task time is not greater than the planned job time, add a timeliness risk mark; Determine the task adaptability according to the safety risk dimension, the maintenance risk dimension and the timeliness risk mark. The task adaptability ranges from level 1 to level 6 from high to low.

[0012] Preferably, the determination of the task adaptability according to the safety risk dimension, the maintenance risk dimension and the timeliness risk mark includes: If there is no safety risk dimension mark, no maintenance risk mark, and no timeliness risk mark, determine the task adaptability as level 1 adaptability; If there is no safety risk dimension mark, no maintenance risk mark, and there is a timeliness risk mark, determine the task adaptability as level 2 adaptability; If there is no safety risk dimension mark, there is one maintenance risk mark, and no timeliness risk mark, determine the task adaptability as level 3 adaptability; If there is no safety risk dimension mark, there is one maintenance risk mark, and there is a timeliness risk mark, determine the task adaptability as level 4 adaptability; If there is no safety risk dimension mark, there are more than one maintenance risk marks, then determine the task adaptability as level 5 adaptability; If there is any safety risk dimension mark, then determine the task adaptability as level 6 adaptability.

[0013] Preferably, the corresponding warning according to the evaluated task adaptability includes: Establish corresponding warning levels according to the task adaptability levels. The higher the adaptability level, the lower the warning level; Adopt different prompting methods for different warning levels, including the display of indicator lights of different colors and the sound and light prompts of different frequencies; Send warning messages to management departments in different scopes according to different warning levels. The higher the warning level, the wider the information sending scope.

[0014] The present invention provides a device warning system based on port digitization, which has the following beneficial effects: By quantitatively processing operation tasks into operation quantization indicators, the system can comprehensively capture the characteristics of operation tasks, including key factors such as the number of containers, weight distribution, and operation space distribution. At the same time, by constructing a deep neural network model, a mapping relationship between operation quantization indicators and device operation characteristics is established, realizing the prediction of device operation characteristics, loss characteristics, and task time-consuming characteristics. This prediction method considering task characteristics significantly improves the prediction accuracy. At the same time, through the prediction correction module, the prediction results are personalized corrected, considering individual differences such as the usage duration and maintenance status of each device, making the prediction results more in line with the actual situation.

[0015] By evaluating the adaptability of the device to the operation task from three dimensions of safety, maintenance requirements, and timeliness, a task adaptability grading mechanism is established. This multi-dimensional and multi-level evaluation mechanism makes risk warning more accurate and targeted, and can clearly distinguish different types and degrees of risks. Furthermore, corresponding warning levels can be established, and risk information is intuitively conveyed through the display of indicator lights of different colors and the sound and light prompts of different frequencies. At the same time, according to different warning levels, the system sends warning messages to management departments in different scopes to ensure that the information can reach personnel with corresponding processing authorities. This hierarchical warning mechanism avoids the unnecessary management burden caused by low-risk events, and at the same time ensures that high-risk events are responded to in a timely and effective manner, significantly improving the efficiency and accuracy of risk disposal. Brief Description of the Drawings

[0016] Figure 1 It is a module schematic diagram of a device warning system based on port digitization of the present invention. Detailed Embodiments

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0018] Embodiment 1 Please refer to Figure 1 In this embodiment, a device warning system based on port digitization includes: A data processing module, which is used to quantitatively process a job task into job quantitative indicators, and is also used to collect job data during the execution of the job task and extract equipment job characteristics; The quantitative processing of the job task into job quantitative indicators includes: Extracting container data indicators, job time indicators, and job space indicators from the job task; The container data indicators include the total number of containers, the total weight of containers, the average weight of containers, the maximum and minimum weights of containers, and the container weight distribution index, which is calculated by the sum of the products of the proportion of the number of containers in each weight interval to the total number and the median of the corresponding interval; the job time indicators include the planned job time; The job space indicators include the maximum and minimum loading and unloading heights, the average loading and unloading height, the loading and unloading height distribution index, the maximum and minimum loading and unloading radii, the average loading and unloading radius, and the loading and unloading radius distribution index, where the loading and unloading height distribution index and the loading and unloading radius distribution index are respectively calculated by the sum of the products of the proportion in each height interval and each radius interval and the median of the corresponding interval; Standardize the container data indicators, job time indicators, and job space indicators to form job quantitative indicators.

[0019] Collecting job data during the execution of the job task and extracting equipment job characteristics includes: Collecting the main motor temperature, hydraulic system pressure value, main structure stress value, and equipment vibration frequency during the execution of the job task, and extracting the maximum values of the above values to form equipment operation characteristics; Collecting the hydraulic oil pollution degree and the wear degree of each connection part during the execution of the job task, and obtaining the corresponding pollution change degree and wear change degree based on the values at the start and end of the job task execution to form equipment loss characteristics; Obtaining the total time taken to complete the job task as the task time characteristic; Using the equipment operation characteristics, equipment loss characteristics, and task time characteristics to form equipment job characteristics.

[0020] In this embodiment, first, the operation task information of the container crane can be obtained through the port production management system port digital system and other methods, and the operation task can be quantified into the operation quantification index. The specific steps of quantification processing can be: extracting container data indicators from the operation task, including the total number of containers, total weight, average weight, maximum and minimum weights, and weight distribution index; the weight distribution index can accurately reflect the distribution characteristics of the container weight in the operation task. At the same time, the planned operation time in the operation time indicator is extracted, which can be used as a reference for the task volume and also as a benchmark value for evaluating the timeliness of the task. For the operation space indicator, the maximum and minimum loading and unloading heights, the average height, the height distribution index, the maximum and minimum loading and unloading radius, the average radius, and the radius distribution index are calculated. The distribution index of loading and unloading height and radius can comprehensively characterize the spatial distribution characteristics of the operation task. Finally, all the extracted indicators are standardized, and the Z-score standardization method can be used to convert indicators of different dimensions into comparable standard scales to form a unified operation quantification indicator vector.

[0021] In terms of equipment operation feature extraction, the data processing module collects real-time data when performing operation tasks through the sensor network arranged on the container crane. First, the operating parameters such as the main motor temperature, hydraulic system pressure value, main structure stress value and equipment vibration frequency are collected, and the maximum values of these parameters in the entire operation process are extracted to form the equipment operation feature vector. These maximum values can reflect the operating status of the equipment and are important indicators for evaluating equipment safety. Secondly, the hydraulic oil contamination and wear data of each connection part are collected. By comparing the measured values at the beginning and after the operation task is completed, the contamination change and wear change are calculated to form the equipment loss feature vector. This before-and-after comparison method can accurately quantify the cumulative loss caused by a single operation to the equipment, providing a direct basis for predicting equipment maintenance needs. Finally, the actual completion time of the operation task is recorded as the task time feature. The equipment operation characteristics, equipment loss characteristics and task time characteristics are combined together to form a complete equipment operation feature vector, which comprehensively reflects the operating status, loss and efficiency performance of the equipment when performing a specific operation task.

[0022] A prediction model module, used to form a training data set using historical operation quantitative indicators and corresponding equipment operation characteristics, and to construct an equipment operation prediction model based on the training data set, wherein the equipment operation prediction model is used to predict the equipment operation characteristics based on the operation quantitative indicators; The constructing of the equipment operation prediction model according to the training data set includes: Perform data cleaning on historical operation quantitative indicators and corresponding equipment operation characteristics, remove outliers and missing values, and form an effective training data set; Divide the effective training data set into a training set and a validation set according to a preset ratio; Construct a deep neural network model structure, including an input layer, multiple hidden layers, and an output layer, where the number of nodes in the input layer is the same as the dimension of the job quantization index, and the number of nodes in the output layer is the same as the dimension of the device job characteristics; Use the job quantization index in the training set as the input and the device job characteristics as the output labels, and adopt the gradient descent optimization algorithm to optimize the model parameters; Through multiple iterative trainings of the training set data, enable the model to learn the mapping relationship between the job quantization index and the device job characteristics; During the training process, use the validation set to evaluate the model performance. When the error value on the validation set no longer decreases after consecutive multiple iterations, terminate the training process; Use the cross-validation method to evaluate the model performance and calculate the error index between the predicted value and the actual value; Adjust the model structure and hyperparameters according to the evaluation results, and repeat the training using the training set until the model performance meets the predetermined threshold requirements to obtain the device job prediction model.

[0023] In this embodiment, first, collect the past operation records of the port container crane, including various job quantization indexes and corresponding device job characteristic data, to form an initial training data set. Clean the historical data to remove outliers and missing values to improve the data quality.

[0024] Randomly divide the effective training data set into a training set and a validation set according to a ratio of 8:2. Next, construct a deep neural network model structure. The model includes an input layer, three hidden layers, and an output layer. The number of nodes in the input layer is the same as the dimension of the job quantization index, which is 15 nodes; the first hidden layer can be set with 64 neurons, the second hidden layer can be set with 128 neurons, and the third hidden layer can be set with 64 neurons; the number of nodes in the output layer is the same as the dimension of the device job characteristics, which is 12 nodes. The ReLU activation function can be adopted for the hidden layers.

[0025] Use the job quantization index in the training set as the input and the device job characteristics as the output labels, and adopt the gradient descent optimization algorithm to optimize the model parameters. Specifically, the Adam optimizer can be used, the initial learning rate is set to 0.001, the batch size is 64, and the mean squared error (MSE) is selected as the loss function. To prevent overfitting, add a Dropout layer after each hidden layer, the dropout rate is set to 0.2, and at the same time, L2 regularization is applied, and the regularization coefficient is 0.001.

[0026] Through multiple iterative trainings of the training set data, the model learns the mapping relationship between the job quantization metrics and the device job characteristics. The maximum number of iterations can be set to 500 rounds, and the performance of the model is evaluated using the validation set after each round of training. When the error value on the validation set no longer decreases for 15 consecutive iterations, the training process is terminated early to avoid overfitting.

[0027] When evaluating the model performance, a 5-fold cross-validation method can be adopted. The effective training data set is evenly divided into 5 parts, and 4 of them are used as training data in turn, and 1 part is used as validation data. Calculate metrics such as the mean squared error, mean absolute error, and R² determination coefficient between the predicted value and the actual value for each validation set, and take the average of the 5 validations as the overall performance evaluation of the model.

[0028] Adjust the model structure and hyperparameters according to the evaluation results, including adjusting the number of hidden layers, the number of neurons in each layer, the learning rate, the regularization coefficient, etc. The system uses a combination of grid search and random search methods to try different hyperparameter combinations and repeats the training using the training set. When the mean squared error of the model on the validation set is lower than the predetermined threshold of 0.05 and the R² determination coefficient is higher than 0.9, it is considered that the model performance meets the requirements, and the final device job prediction model is obtained.

[0029] This prediction model based on deep learning can capture the complex non-linear relationship between the job quantization metrics and the device job characteristics, and improves the prediction accuracy compared with traditional statistical methods. By learning the patterns in a large amount of historical data, the model can accurately predict the operating load and loss conditions caused by different job tasks to the device, providing a reliable prediction basis for subsequent task adaptability evaluation.

[0030] A real-time prediction module for obtaining the job quantization metrics corresponding to the next job task to be executed and predicting the device job characteristics through the device job prediction model; In this embodiment, when a new job task is assigned to the container crane, the real-time prediction module can process the job task into a standardized job quantization metric vector through the data processing module. Subsequently, the real-time prediction module inputs these job quantization metrics into the trained device job prediction model, and the model automatically calculates and outputs the predicted device job characteristics according to the input metrics. This real-time prediction ability enables the system to evaluate the potential impact of the task on the device before the job starts, providing data support for subsequent risk warnings.

[0031] A prediction correction module for obtaining a performance deviation vector according to the device job characteristics predicted by the device job prediction model and the corresponding actually extracted device job characteristics of the execution device in history; correcting the predicted device job characteristics according to the performance deviation vector to obtain corrected job characteristic data; The steps for obtaining the performance deviation vector include: Collect the predicted device operation characteristics and the actually extracted device operation characteristic data in at least the last five operations of the execution device; For the predicted and actual characteristics of each operation, calculate the difference between the actual value and the predicted value for each dimension; Arrange the differences of each dimension in chronological order to form a deviation sequence for each dimension; Predict the expected deviation value of each current dimension based on the deviation sequences of each dimension; Combine the expected deviation values of each dimension to form a performance deviation vector.

[0032] The predicting the expected deviation value of each current dimension based on the deviation sequences of each dimension includes: Perform outlier detection and elimination on the deviation sequences of each dimension; For the deviation sequences of each dimension after eliminating outliers, calculate the weighted mean, deviation change rate, and fluctuation index; Taking the most recent deviation value in the deviation sequence as a benchmark, obtain a preliminary deviation by combining the change rate and the time interval; Judge the fluctuation index. If the fluctuation index is lower than the set threshold, directly use the preliminary deviation as the expected deviation value; Otherwise, fuse the weighted mean and the preliminary deviation, and take their arithmetic mean as the expected deviation value.

[0033] In this embodiment, although the deep learning model can establish a mapping relationship between the operation quantization index and the device operation characteristics, due to factors such as the usage duration and maintenance status of each device, there are often certain deviations between the prediction results and the actual situation. The prediction correction module analyzes the historical prediction deviation patterns of specific devices to perform personalized correction on the prediction results, improving the prediction accuracy. The specific modification process can be as follows: First, collect the recent operation records of the execution device, including at least the predicted and actual characteristic data of five complete operations. Taking a container crane as an example, the system can record the predicted and actual maximum values of the main motor temperature of the device in its last five operations, calculate the differences of each operation to form a deviation sequence. Perform outlier detection on the deviation sequence. The modified Z-score method can be used, and the median absolute deviation (MAD) is calculated to identify outliers. Set an appropriate outlier threshold. When a certain deviation value exceeds the threshold range, it is regarded as an outlier and removed from the sequence to ensure the accuracy of subsequent analysis.

[0034] Next, calculate the weighted mean, deviation change rate, and fluctuation index. For the weighted mean, a time decay weight method can be adopted to assign higher weights to the most recent deviation values, reflecting the importance of recent data for the current prediction. The deviation change rate can be calculated by a linear regression method to obtain the overall change trend of the deviation sequence and the average deviation growth value for each operation. The fluctuation index can be obtained by calculating the ratio of the standard deviation to the mean value, which is used to evaluate the stability of the deviation sequence. The system can preset a fluctuation threshold for subsequent decision-making.

[0035] Based on the most recent deviation value, combined with the calculated change rate and time interval (such as the number of operation cycles), the preliminary deviation value of the current operation can be deduced. Subsequently, the system determines whether the fluctuation index is lower than the preset threshold. If the fluctuation index is low, indicating that the deviation change is relatively stable, the preliminary deviation can be directly used as the expected deviation value; if the fluctuation index is high, indicating significant deviation fluctuations, the weighted mean and the preliminary deviation can be combined, and their arithmetic mean can be taken as a more robust expected deviation value. The same calculation process is performed for all dimensions of the equipment operation characteristics, and finally, a complete performance deviation vector is formed.

[0036] Apply the performance deviation vector to the prediction result of the current task for dimension - corresponding correction. For example, add the predicted maximum temperature of the main motor to the corresponding deviation value to obtain the corrected temperature value. The same correction is made for other dimensions, and finally, complete corrected operation characteristic data is obtained.

[0037] By correcting the equipment operation characteristics predicted by the model in this way, since the personalized operation characteristics of the equipment are captured, the corrected equipment operation characteristics are more in line with the actual situation. At the same time, this method does not require retraining the prediction model, has high computational efficiency, and is suitable for real - time application scenarios.

[0038] A task evaluation module is used to evaluate the task adaptability of the executing equipment for the next operation task to be executed according to the corrected operation characteristic data and operation quantification indicators; The evaluation of the task adaptability of the executing equipment for the next operation task to be executed includes: Compare each dimension value of the equipment operation characteristics in the corrected operation characteristic data with the corresponding safe operation threshold respectively; mark the dimensions that exceed the corresponding safety threshold as safety risk dimensions; Obtain the current hydraulic oil pollution degree and wear degree data of each connection part of the executing equipment, and add them to the corresponding pollution change degree and wear change degree in the corrected operation characteristic data respectively to obtain the predicted pollution degree and predicted wear degree; Compare the predicted pollution degree and predicted wear degree with the corresponding maintenance thresholds respectively, and mark the dimensions that exceed the corresponding maintenance thresholds as maintenance risk dimensions; Compare the task time feature in the corrected operation feature data with the operation time index. If the task time is not greater than the planned operation time, add a timeliness risk mark. Determine the task adaptability according to the safety risk dimension, maintenance risk dimension, and timeliness risk mark. The task adaptability ranges from level 1 to level 6 from high to low.

[0039] The determination of the task adaptability according to the safety risk dimension, maintenance risk dimension, and timeliness risk mark includes: If there is no safety risk dimension mark, no maintenance risk mark, and no timeliness risk mark, determine the task adaptability as level 1 adaptability; If there is no safety risk dimension mark, no maintenance risk mark, and there is a timeliness risk mark, determine the task adaptability as level 2 adaptability; If there is no safety risk dimension mark, there is one maintenance risk mark, and no timeliness risk mark, determine the task adaptability as level 3 adaptability; If there is no safety risk dimension mark, there is one maintenance risk mark, and there is a timeliness risk mark, determine the task adaptability as level 4 adaptability; If there is no safety risk dimension mark, there are more than one maintenance risk marks, determine the task adaptability as level 5 adaptability; If there is any safety risk dimension mark, determine the task adaptability as level 6 adaptability.

[0040] In this embodiment, the task evaluation module evaluates the adaptability of the device to perform the next operation task from three dimensions of safety, maintenance requirements, and timeliness by comparing the corrected operation feature data with various thresholds. The various thresholds can be preset by comprehensively considering factors such as the specific model and service life of the device.

[0041] By comparing the device operation characteristics with the safety threshold, predicting the pollution degree and wear degree with the maintenance threshold, and comparing the task time with the planned time, the system can comprehensively evaluate the possible risk points in the task execution process. This data-based evaluation method eliminates the subjectivity of traditional experience judgment and makes the task adaptability evaluation more scientific.

[0042] At the same time, the hierarchical design provides structured input data for the subsequent alarm module, enabling the early warning information to accurately reflect the risk type and severity. Furthermore, different staff can timely decide whether to adjust the task assignment, replace the execution device, or arrange device maintenance according to the adaptability results of different levels.

[0043] The alarm module is used to perform corresponding early warnings according to the evaluated task adaptability.

[0044] The corresponding early warning according to the evaluated task adaptability includes: Establish corresponding warning levels according to the task adaptability level, where the higher the adaptability level, the lower the warning level; Adopt different prompting methods for different warning levels, including indicating lights of different colors and audible and visual prompts of different frequencies; Send warning messages to management departments in different scopes for different warning levels. The higher the warning level, the wider the information sending scope.

[0045] In this embodiment, the alarm module establishes a corresponding warning mechanism according to the task adaptability level output by the task evaluation module to achieve timely and effective transmission of risks. The warning level is inversely proportional to the task adaptability level. The higher the task adaptability level (the lower the risk), the lower the warning level; the lower the task adaptability level (the higher the risk), the higher the warning level.

[0046] Transmit warning information through multiple sensory methods, including visual and audible prompts. For visual prompts, adopt an intuitive color coding system. For example, level 1 adaptability corresponds to a green indicating light, indicating safety and no risk; level 2 adaptability corresponds to a blue indicating light, indicating a slight failure risk; levels 3 and 4 adaptability correspond to a yellow indicating light, indicating a maintenance risk that requires attention; level 5 adaptability corresponds to an orange indicating light, indicating multiple maintenance risks; level 6 adaptability corresponds to a red indicating light, indicating a safety risk. The frequency of the audible and visual prompts also increases with the increase of the warning level, from no prompt to continuous high-frequency alarm, enabling the operator to quickly perceive the risk level.

[0047] The transmission scope of the warning information is reasonably extended according to the risk level. For example, low-level warnings (such as levels 1-2) only notify the equipment operator and the direct manager; medium-level warnings (such as levels 3-4) are extended to the maintenance department and the dispatching center; high-level warnings (such as levels 5-6) are reported to the safety management department and the port management layer. This hierarchical transmission mechanism ensures that information can reach personnel with corresponding processing authorities, avoids unnecessary management burdens caused by low-risk events, and at the same time ensures that high-risk events receive timely and effective responses.

[0048] This warning mechanism converts the complex risk assessment results into simple and clear warning signals through a standardized information transmission method, facilitating quick understanding and response by personnel at all levels.

[0049] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only one type, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0050] As described above, the above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

[0051] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should all be included in the protection scope of the present invention.

Claims

1. An equipment warning system based on port digitization, characterized in that, Including: A data processing module, which is used to quantitatively process a job task into job quantization metrics, and is also used to collect job data during the execution of the job task and extract device job characteristics; A prediction model module, which is used to use historical job quantization metrics and corresponding device job characteristics to form a training data set, and construct a device job prediction model based on the training data set. The device job prediction model is used to predict device job characteristics according to job quantization metrics; A real-time prediction module, which is used to obtain job quantization metrics corresponding to the next job task to be executed, and predict device job characteristics through the device job prediction model; A prediction correction module, which is used to obtain a performance deviation vector according to the device job characteristics predicted by the device job prediction model in history and the corresponding actually extracted device job characteristics of the execution device; correct the predicted device job characteristics according to the performance deviation vector to obtain corrected job characteristic data; A task evaluation module, which is used to evaluate the task adaptability of the execution device to the next job task to be executed according to the corrected job characteristic data and job quantization metrics; An alarm module, which is used to give corresponding early warnings according to the evaluated task adaptability.

2. The equipment warning system based on port digitization according to claim 1, characterized in that, The quantitative processing of the job task into job quantization metrics includes: Extracting container data metrics, job time metrics, and job space metrics from the job task; The container data metrics include the total number of containers, the total weight of containers, the average weight of containers, the maximum and minimum weights of containers, and the container weight distribution index. The container weight distribution index is calculated by the sum of the products of the proportion of the number of containers in each weight interval to the total number and the median of the corresponding interval; the job time metrics include the planned job time; The job space metrics include the maximum and minimum loading and unloading heights, the average loading and unloading height, the loading and unloading height distribution index, the maximum and minimum loading and unloading radii, the average loading and unloading radius, and the loading and unloading radius distribution index. The loading and unloading height distribution index and the loading and unloading radius distribution index are respectively calculated by the sum of the products of the proportion of each height interval and each radius interval and the median of the corresponding interval; Standardize the container data metrics, job time metrics, and job space metrics to form job quantization metrics.

3. The device warning system based on port digitization according to claim 2, characterized in that, The collection of job data during the execution of the job task and the extraction of device job characteristics include: Collect the main motor temperature, hydraulic system pressure value, main structure stress value, and device vibration frequency during the execution of the job task, and extract the maximum values of the above values to form device operation characteristics; Collect the hydraulic oil pollution degree and the wear degree of each connection part during the execution of the job task, and obtain the corresponding pollution change degree and wear change degree according to the values at the start and end of the job task to form device loss characteristics; Obtain the total time used to complete the job task as the task time characteristic; Use the device operation characteristics, device loss characteristics, and task time characteristics to form device job characteristics.

4. The device warning system based on port digitization according to claim 3, characterized in that, The construction of the device job prediction model according to the training data set includes: Perform data cleaning on historical job quantization metrics and corresponding device job characteristics, remove outliers and missing values, and form an effective training data set; Divide the effective training data set into a training set and a validation set according to a preset ratio; Construct a deep neural network model structure, including an input layer, multiple hidden layers, and an output layer, where the number of nodes in the input layer is the same as the dimension of the job quantization index, and the number of nodes in the output layer is the same as the dimension of the device job characteristics; Use the job quantization index in the training set as the input and the device job characteristics as the output label, and use the gradient descent optimization algorithm to optimize the model parameters; Through multiple iterative trainings of the training set data, enable the model to learn the mapping relationship between the job quantization index and the device job characteristics; During the training process, use the validation set to evaluate the model performance. When the error value on the validation set no longer decreases for multiple consecutive iterations, terminate the training process; Use the cross-validation method to evaluate the model performance and calculate the error index between the predicted value and the actual value; Adjust the model structure and hyperparameters according to the evaluation results, and repeat the training using the training set until the model performance meets the predetermined threshold requirements to obtain a device job prediction model.

5. The device warning system based on port digitization according to claim 4, wherein The steps for obtaining the performance deviation vector include: Collect the predicted device job characteristics and the actually extracted device job characteristic data in at least the last five jobs executed by the execution device; For the predicted and actual characteristics of each job, calculate the difference between the actual value and the predicted value for each dimension; Arrange the differences of each dimension in chronological order to form a deviation sequence for each dimension; Predict the expected deviation value of each current dimension according to the deviation sequence of each dimension; Combine the expected deviation values of each dimension to form a performance deviation vector.

6. The device warning system based on port digitization according to claim 5, characterized in that, The predicting the expected deviation value of each current dimension according to the deviation sequence of each dimension includes: Perform outlier detection and removal on the deviation sequence of each dimension; For the deviation sequence of each dimension after removing outliers, calculate the weighted mean, deviation change rate, and fluctuation index; Based on the most recent deviation value in the deviation sequence, combine the change rate and the time interval to obtain a preliminary deviation; Judge the fluctuation index. If the fluctuation index is lower than the set threshold, directly use the preliminary deviation as the expected deviation value; Otherwise, fuse the weighted mean and the preliminary deviation, and take their arithmetic mean as the expected deviation value.

7. The device warning system based on port digitization according to claim 6, characterized in that, The evaluating the task adaptability of the execution device for the next job task to be executed includes: Compare the values of each dimension of the device operation characteristics in the corrected job characteristic data with the corresponding safe operation threshold respectively; Mark the dimensions exceeding the corresponding safety threshold as safety risk dimensions; Obtain the current hydraulic oil pollution degree and the wear degree data of each connection part of the execution device, and add them to the corresponding pollution change degree and wear change degree in the corrected job characteristic data respectively to obtain the predicted pollution degree and the predicted wear degree; Compare the predicted pollution degree and the predicted wear degree with the corresponding maintenance thresholds respectively, and mark the dimensions exceeding the corresponding maintenance thresholds as maintenance risk dimensions; Compare the task time-consuming characteristic in the corrected job characteristic data with the job time index. If the task time-consuming is not greater than the planned job time-consuming, add a timeliness risk mark; Determine the task adaptability according to the safety risk dimension, the maintenance risk dimension, and the timeliness risk mark. The task adaptability ranges from 1 to 6 levels from high to low.

8. The device warning system based on port digitization according to claim 7, characterized in that, The determination of task adaptability based on the safety risk dimension, maintenance risk dimension, and timeliness risk marker includes: If there is no safety risk dimension marker, no maintenance risk marker, and no timeliness risk marker, the task adaptability is determined as level 1 adaptability; If there is no safety risk dimension marker, no maintenance risk marker, and there is a timeliness risk marker, the task adaptability is determined as level 2 adaptability; If there is no safety risk dimension marker, there is one maintenance risk marker, and no timeliness risk marker, the task adaptability is determined as level 3 adaptability; If there is no safety risk dimension marker, there is one maintenance risk marker, and there is a timeliness risk marker, the task adaptability is determined as level 4 adaptability; If there is no safety risk dimension marker, there are more than one maintenance risk markers, the task adaptability is determined as level 5 adaptability; If there is any safety risk dimension marker, the task adaptability is determined as level 6 adaptability.

9. The device warning system based on port digitization according to claim 8, wherein The corresponding early warning based on the evaluated task adaptability includes: Establish corresponding early warning levels according to the task adaptability level, and the higher the adaptability level, the lower the early warning level; Adopt different prompting methods for different early warning levels, including the display of indicator lights of different colors and the acoustic-optical prompts of different frequencies; Send early warning information to management departments in different scopes for different early warning levels. The higher the early warning level, the wider the information sending scope.

Images