Salt pan production equipment predictive maintenance system based on artificial intelligence

Abstract

The invention relates to the technical field of salt pan production, and discloses a salt pan production equipment predictive maintenance system based on artificial intelligence, and the system comprises an equipment real-time monitoring and dynamic maintenance period modeling module which obtains the real-time operation data of salt pan production equipment and builds a dynamic maintenance period model; the equipment abnormal state trend prediction module is used for carrying out abnormal state development trend prediction based on the dynamic maintenance period model and equipment historical fault features; and the key equipment health state comprehensive evaluation and priority ranking module is used for carrying out comprehensive evaluation and priority ranking on the health state of the key equipment according to the multi-source sensing data. According to the salt pan production equipment predictive maintenance system based on artificial intelligence, by obtaining the real-time operation data of the salt pan production equipment and establishing the dynamic maintenance period model, dynamic regulation and control of the maintenance period are achieved, and the problems of resource waste and equipment failure rate increase caused by over-maintenance or under-maintenance are effectively avoided.

Description

Technical Field

[0001] This application relates to the field of salt field production technology, and in particular to a predictive maintenance system for salt field production equipment based on artificial intelligence. Background Technology

[0002] Our predictive maintenance system for salt field production equipment uses artificial intelligence technology to "diagnose" the equipment's operation and maintenance. Simply put, it relies on big data analysis and machine learning algorithms to monitor the equipment's operating status in real time and intelligently determine if there are any problems. This improves the equipment's reliability and makes maintenance much easier.

[0003] However, to truly perfect this system, several real challenges need to be addressed. For example, how to determine the appropriate maintenance cycle? It can't be static; it needs to be flexibly adjusted based on real-time equipment operating data. Otherwise, either too much maintenance leads to waste, or insufficient maintenance causes problems. Also, how can we detect impending equipment failure in advance? This requires combining past failure scenarios with the current operating status to anticipate the development of anomalies, avoiding panic and chaos after a sudden shutdown.

[0004] Furthermore, the health status of equipment cannot be determined by looking at a single data point. Information collected from various sensors must be aggregated and analyzed to accurately assess the equipment's condition and, consequently, to rationally prioritize maintenance.

[0005] Salt field production is also greatly affected by climate; changes in evaporation rate necessitate adjustments to equipment load. The system must be able to automatically optimize equipment operating load based on changes in climate conditions to ensure undiminished operational efficiency.

[0006] Finally, spare parts inventory is also a headache. Too much inventory takes up space and wastes money, while too little will not be able to keep up if the equipment breaks down. The system needs to be able to intelligently adjust spare parts inventory according to the degradation trend of the equipment, so as to avoid both backlog and shortage. Only in this way can the entire maintenance strategy be more accurate, more efficient and more sustainable. Summary of the Invention

[0007] To address the problems mentioned in the background, this application provides a predictive maintenance system for salt field production equipment based on artificial intelligence.

[0008] This application provides a predictive maintenance system for salt field production equipment based on artificial intelligence, employing the following technical solution:

[0009] An artificial intelligence-based predictive maintenance system for salt field production equipment includes:

[0010] The real-time equipment monitoring and dynamic maintenance cycle modeling module acquires real-time operating data of salt field production equipment and establishes a dynamic maintenance cycle model.

[0011] The equipment abnormal state trend prediction module predicts the development trend of abnormal states based on the dynamic maintenance cycle model and the historical fault characteristics of the equipment.

[0012] The critical equipment health status comprehensive assessment and priority ranking module comprehensively assesses and prioritizes the health status of critical equipment based on multi-source sensor data.

[0013] The intelligent optimization and control module for spare parts inventory management strategies intelligently optimizes and controls spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and changes in evaporation rates output by the predictive model.

[0014] Preferably, the step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0015] Collect equipment vibration spectrum data, oil contamination index, current fluctuation parameters, and temperature change curves;

[0016] Calculate the operational stability index of the equipment ,in The standard deviation of the measurement data This represents the mean;

[0017] By combining historical maintenance records with current operating data, a dynamic maintenance cycle adjustment coefficient is generated. ,in It is the equipment load factor. It's about operational efficiency;

[0018] The adjustment factor K is used to determine whether to initiate a new maintenance cycle prediction model for optimization.

[0019] Preferably, the prediction of abnormal state development trends based on the dynamic maintenance cycle model and historical equipment fault characteristics further includes:

[0020] Collect the frequency of failures in historical failure modes Mean Time Between Failures and equipment degradation rate ;

[0021] Calculate the probability of anomalies ,in This is the cumulative time since the last maintenance;

[0022] A time-series trend prediction network is constructed using an LSTM model to perform multi-step prediction of abnormal development paths;

[0023] By setting a threshold Determine whether an early warning has been triggered. A high-risk warning is given; otherwise, a low-risk prediction is made.

[0024] The comprehensive assessment and prioritization of the health status of key equipment based on multi-source sensor data further includes:

[0025] Extract data from pressure sensors, flow meters, vibration detectors, and temperature probes;

[0026] Calculate the health index of each device (weight) Signal fluctuation value ); where weight It is the preset equipment importance coefficient;

[0027] According to health index Rank the equipment and schedule maintenance according to the ranking order;

[0028] Clustering algorithms are used to divide devices into normal, early warning, and emergency groups for subsequent resource allocation.

[0029] Preferably, the intelligent optimization and control of spare parts inventory management strategy based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes:

[0030] Extract the degradation rate predicted by the model And the actual monitored evaporation rate of salt fields ;

[0031] Calculate demand volatility ,in This represents the cumulative historical values ​​over the past 7 days;

[0032] An improved Markov model is used to simulate inventory demand, and the spare parts turnover rate is defined. ;

[0033] When spare parts turnover rate Greater than or equal to the threshold A dynamic adjustment mechanism for vehicle inventory.

[0034] Preferably, the step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0035] Load rate of collection equipment: Load = (Operating energy consumption / Rated power) ;

[0036] Introducing seasonal adjustment coefficient ,in This is the average temperature over the past week. For reference temperature;

[0037] Adjusting maintenance cycles using a normalization formula: CT ,in This is the basic maintenance cycle. This is an empirical coefficient;

[0038] Set the minimum maintenance cycle Tmin=7 days to prevent over-maintenance.

[0039] Preferably, the prediction of abnormal state development trends based on the dynamic maintenance cycle model and historical equipment fault characteristics further includes:

[0040] Construct a fuzzy inference system, with input variables including device operating time. Number of abnormal features Maintenance interval ;

[0041] The output variable is the device failure rate. ;

[0042] The prediction strength of different levels of faults is defined using a fuzzy rule set;

[0043] Based on expert experience, the prediction results are mapped to tiered alarms (red / yellow / green) to guide response strategies.

[0044] Preferably, the step of comprehensively assessing and prioritizing the health status of key equipment based on multi-source sensor data further includes:

[0045] Principal component analysis (PCA) is used to extract common feature components from each sensor data.

[0046] Define the health weight matrix ,in It is the sensor sensitivity coefficient. It is the reading intensity. It is the exponential coefficient;

[0047] Based on total weight Classify the equipment;

[0048] Set maintenance priority criteria: If If it is a high level, then it is a high level. That would be a low level.

[0049] Preferably, the intelligent optimization and control of spare parts inventory management strategy based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes:

[0050] Simulate future weather using historical climate models and estimate potential impacts;

[0051] Introducing the correlation coefficient of evaporation rate ;

[0052] Calculate the safety stock of spare parts ,in It is the spare parts shortage tolerance factor. It predicts the frequency of failures;

[0053] Set upper and lower limits for inventory control. When a supplementary instruction is issued, Adjust inventory strategies as needed.

[0054] Preferably, the step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0055] Analyze the vibration frequency domain characteristics of the equipment and identify the resonant frequency band. ;

[0056] Calculating spectral energy density using a spectrogram Within the range;

[0057] Calculate the operational stability index of the equipment ;

[0058] Combined with equipment aging coefficient Establish a dynamic maintenance cycle model ,in This is the basic maintenance cycle.

[0059] Preferably, the intelligent optimization and control of spare parts inventory management strategy based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes:

[0060] Multimodal fusion learning of device image features and vibration data is performed using a convolutional neural network (CNN);

[0061] Reinforcement learning models are trained based on historical inventory and scheduling data to optimize replenishment strategies;

[0062] The optimal strategy is generated using the Q-learning algorithm, and a reward function is defined. ;

[0063] The model is updated based on the reward value after each decision to achieve adaptive inventory management.

[0064] In summary, this application includes at least one of the following beneficial technical effects:

[0065] 1. This AI-based predictive maintenance system for salt field production equipment acquires real-time operating data of the equipment and establishes a dynamic maintenance cycle model, thereby achieving dynamic control of the maintenance cycle and effectively avoiding resource waste and increased equipment failure rates caused by over-maintenance or under-maintenance.

[0066] 2. This AI-based predictive maintenance system for salt field production equipment predicts the development trend of abnormal states based on a dynamic maintenance cycle model and historical equipment fault characteristics. It can detect potential equipment abnormalities in advance, thereby reducing production interruptions caused by sudden equipment shutdowns.

[0067] 3. This AI-based predictive maintenance system for salt field production equipment comprehensively assesses and prioritizes the health status of key equipment based on multi-source sensor data, which helps improve the efficiency of maintenance decisions and makes resource allocation more rational, avoiding ineffective resource occupation or shortage.

[0068] 4. This AI-based predictive maintenance system for salt field production equipment adaptively adjusts the load on key equipment by combining climate conditions and the changing trend of salt field evaporation rate, thus solving the problem of shortened equipment lifespan due to overload operation or production impact due to low efficiency.

[0069] 5. This AI-based predictive maintenance system for salt field production equipment intelligently optimizes and controls spare parts inventory management strategies based on the equipment degradation trends output by the predictive model. This avoids maintenance delays caused by spare parts shortages or capital tied up in spare parts, and improves the timeliness of maintenance response. Attached Figure Description

[0070] Figure 1 This is a flowchart of an AI-based predictive maintenance system for salt field production equipment. Detailed Implementation

[0071] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0072] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0073] This application discloses an artificial intelligence-based predictive maintenance system for salt field production equipment, as shown in the attached document. Figure 1 This describes an artificial intelligence-based predictive maintenance system for salt field production equipment, including:

[0074] The real-time equipment monitoring and dynamic maintenance cycle modeling module acquires real-time operating data of salt field production equipment and establishes a dynamic maintenance cycle model.

[0075] The equipment abnormal state trend prediction module predicts the development trend of abnormal states based on the dynamic maintenance cycle model and the historical fault characteristics of the equipment.

[0076] The critical equipment health status comprehensive assessment and priority ranking module comprehensively assesses and prioritizes the health status of critical equipment based on multi-source sensor data.

[0077] The intelligent optimization and control module for spare parts inventory management strategies intelligently optimizes and controls spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and changes in evaporation rates output by the predictive model.

[0078] Acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0079] Collect equipment vibration spectrum data, oil contamination index, current fluctuation parameters, and temperature change curves;

[0080] Calculate the operational stability index of the equipment ,in The standard deviation of the measurement data This represents the mean;

[0081] By combining historical maintenance records with current operating data, a dynamic maintenance cycle adjustment coefficient is generated. ,in It is the equipment load factor. It's about operational efficiency;

[0082] The adjustment factor K is used to determine whether to initiate a new maintenance cycle prediction model for optimization.

[0083] The prediction of abnormal state development trends based on dynamic maintenance cycle models and historical equipment fault characteristics further includes:

[0084] Collect the frequency of failures in historical failure modes Mean Time Between Failures and equipment degradation rate ;

[0085] Calculate the probability of anomalies ,in This is the cumulative time since the last maintenance;

[0086] A time-series trend prediction network is constructed using an LSTM model to perform multi-step prediction of abnormal development paths;

[0087] By setting a threshold Determine whether an early warning has been triggered. A high-risk warning is given; otherwise, a low-risk prediction is made.

[0088] The comprehensive assessment and prioritization of the health status of key equipment based on multi-source sensor data further includes:

[0089] Extract data from pressure sensors, flow meters, vibration detectors, and temperature probes;

[0090] Calculate the health index of each device (weight) Signal fluctuation value ); where weight It is the preset equipment importance coefficient;

[0091] According to health index Rank the equipment and schedule maintenance according to the ranking order;

[0092] Clustering algorithms are used to divide devices into normal, early warning, and emergency groups for subsequent resource allocation.

[0093] The intelligent optimization and control of spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and evaporation rate changes output by the predictive model further includes:

[0094] Extract the degradation rate predicted by the model And the actual monitored evaporation rate of salt fields ;

[0095] Calculate demand volatility ,in This represents the cumulative historical values ​​over the past 7 days;

[0096] An improved Markov model is used to simulate inventory demand, and the spare parts turnover rate is defined. ;

[0097] When spare parts turnover rate Greater than or equal to the threshold A dynamic adjustment mechanism for vehicle inventory.

[0098] Acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0099] Load rate of collection equipment: Load = (Operating energy consumption / Rated power) ;

[0100] Introducing seasonal adjustment coefficient ,in This is the average temperature over the past week. For reference temperature;

[0101] Adjusting maintenance cycles using a normalization formula: CT ,in This is the basic maintenance cycle. This is an empirical coefficient;

[0102] Set the minimum maintenance cycle Tmin=7 days to prevent over-maintenance.

[0103] The prediction of abnormal state development trends based on dynamic maintenance cycle models and historical equipment fault characteristics further includes:

[0104] Construct a fuzzy inference system, with input variables including device operating time. Number of abnormal features Maintenance interval ;

[0105] The output variable is the device failure rate. ;

[0106] The prediction strength of different levels of faults is defined using a fuzzy rule set;

[0107] Based on expert experience, the prediction results are mapped to tiered alarms (red / yellow / green) to guide response strategies.

[0108] The comprehensive assessment and prioritization of the health status of key equipment based on multi-source sensor data further includes:

[0109] Principal component analysis (PCA) is used to extract common feature components from each sensor data.

[0110] Define the health weight matrix ,in It is the sensor sensitivity coefficient. It is the reading intensity. It is the exponential coefficient;

[0111] Based on total weight Classify the equipment;

[0112] Set maintenance priority criteria: If If it is a high level, then it is a high level. That would be a low level.

[0113] The intelligent optimization and control of spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and evaporation rate changes output by the predictive model further includes:

[0114] Simulate future weather using historical climate models and estimate potential impacts;

[0115] Introducing the correlation coefficient of evaporation rate ;

[0116] Calculate the safety stock of spare parts ,in It is the spare parts shortage tolerance factor. It predicts the frequency of failures;

[0117] Set upper and lower limits for inventory control. When a supplementary instruction is issued, Adjust inventory strategies as needed.

[0118] Acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes:

[0119] Analyze the vibration frequency domain characteristics of the equipment and identify the resonant frequency band. ;

[0120] Calculating spectral energy density using a spectrogram Within the range;

[0121] Calculate the operational stability index of the equipment ;

[0122] Combined with equipment aging coefficient Establish a dynamic maintenance cycle model ,in This is the basic maintenance cycle.

[0123] The intelligent optimization and control of spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and evaporation rate changes output by the predictive model further includes:

[0124] Multimodal fusion learning of device image features and vibration data is performed using a convolutional neural network (CNN);

[0125] Reinforcement learning models are trained based on historical inventory and scheduling data to optimize replenishment strategies;

[0126] The optimal strategy is generated using the Q-learning algorithm, and a reward function is defined. ;

[0127] The model is updated based on the reward value after each decision to achieve adaptive inventory management.

[0128] The system comprises several steps: First, it collects equipment vibration spectrum data, oil contamination index, current fluctuation parameters, and temperature change curves. This data reflects the equipment's operating status and potential failure risks, aiding in subsequent analysis. For example, in salt field production equipment, real-time collection of centrifugal pump vibration data via sensors can reveal bearing wear trends. The oil contamination index, through periodic sampling, checks the cleanliness of the lubrication system, preventing equipment damage due to poor lubrication. Current fluctuation parameters indicate whether the motor load is within the normal range, while the temperature change curve helps determine if there is abnormal heating.

[0129] Secondly, calculate the operational stability index of the equipment. ,in The standard deviation of the measurement data This represents the mean. This formula measures the stability of equipment operation by the ratio of the standardized variance to the mean. The smaller the value, the more concentrated the data. The larger the value, the smoother the operation. A higher value indicates greater device stability. In practical applications, if... A value >0.8 indicates that the system is running stably and does not require frequent maintenance.

[0130] Subsequently, by combining historical maintenance records with current operating data, a dynamic maintenance cycle adjustment coefficient is generated. ,in It is the equipment load factor, which is usually between 0 and 1. This refers to operating efficiency, with a value ranging from 0 to 1. and These are weighting factors used to balance the effects of load and efficiency. For example, when equipment operates at full load for an extended period and efficiency decreases, The value will increase, triggering more frequent maintenance reminders.

[0131] Finally, based on the adjustment coefficient Determine whether to initiate a new maintenance cycle prediction model for optimization. If If the value exceeds a preset threshold, the maintenance plan will be updated to improve equipment reliability and extend its service life. This solution enables more intelligent, efficient, and personalized maintenance management, reducing downtime and maintenance costs.

[0132] In this embodiment, the system is deployed at a salt field production site, and the main monitoring targets include key production equipment such as centrifugal pumps, evaporation tank agitators, conveyor belt motors, and brine filtration equipment. The system achieves predictive maintenance throughout the entire process through the following steps:

[0133] 1. Multi-source heterogeneous data acquisition and fusion

[0134] The system collects the following multimodal data in real time through an IoT sensor network deployed at key parts of the equipment:

[0135] Vibration data: A triaxial accelerometer (such as ADI ADXL355) was used with a sampling frequency of 10 kHz to collect vibration acceleration signals in the X, Y, and Z directions of the device.

[0136] Oil data: Online oil quality sensors (such as Parker Kittiwake FG) monitor the changes in water content, particulate matter concentration and viscosity in lubricating oil in real time, and generate the oil contamination index (OCI).

[0137] Electrical parameters: The three-phase current of the motor is collected by clamp current sensors (such as Fluke i400), and the current imbalance and harmonic distortion rate are calculated.

[0138] Temperature data: The PT100 resistance temperature detector monitors the temperature of key components such as bearings and windings, and generates temperature change curves.

[0139] All sensor data is uploaded via industrial IoT gateways (such as Huawei AR502H) using the MQTT protocol to edge computing nodes (such as Intel NUC) for preliminary filtering and feature extraction, and then transmitted to the cloud analysis platform.

[0140] 2. Specific implementation of dynamic maintenance cycle modeling

[0141] The system employs a sliding time window (24 hours) to perform a short-time Fourier transform (STFT) on the acquired vibration spectrum data to extract frequency band energy characteristics. The formula for calculating the operational stability index SS is modified as follows:

[0142]

[0143] in To prevent extremely small constants from being divided by zero, an environmental corrosion factor is introduced into the system to address the high humidity and high corrosion environment of actual salt fields. This factor is derived by monitoring the chloride ion concentration in the air using a salt spray sensor and combining it with a corrosion rate model of the equipment material. The adjusted dynamic dimension... Expanded to:

[0144]

[0145] in This is the corrosion weighting coefficient, which is dynamically set according to the corrosion level of the salt field area where the equipment is located.

[0146] 3. Training of Deep Learning Models for Anomaly Trend Prediction

[0147] Historical fault data was obtained from the salt field's work order records and SCADA system alarm logs over the past three years, collecting a total of 127 valid fault events. The system uses a model combining a Temporal Convolutional Network (TCN) and an attention mechanism to predict anomaly development. The network structure is as follows:

[0148] Input layer: Multi-dimensional feature sequences of the past 30 days (vibration energy, temperature gradient, current harmonics, etc.).

[0149] TCN layer: convolution kernel size 7, dilation coefficients of 1, 2, 4, 8, extracting multi-scale temporal features.

[0150] Attention layer: Give higher weight to key fault symptoms (such as a sudden increase in vibration energy in a specific frequency band).

[0151] Output layer: Anomaly probability curves for each device over the next 7 days.

[0152] The model uses five-fold cross-validation during training, and the AUC on the final test set reaches 0.92, which is significantly better than the traditional LSTM model.

[0153] 4. Adaptive Clustering for Health Status Assessment and Prioritization

[0154] The system uses a Gaussian Mixture Model (GMM) to cluster the equipment health index HIHI, replacing the original K-means algorithm, to adapt to the non-linear characteristics of the health status distribution. In addition to the three groups of "normal, warning, and emergency", a new "observation group" has been added to the clustering results. This group corresponds to equipment with a moderate but fluctuating health index, for which it is recommended to increase the frequency of inspections.

[0155] 5. Optimization of Salt Field Climate Adaptation and Inventory Linkage

[0156] The system integrates local weather station data with NOAA climate prediction data to establish evaporation rate. With equipment load rate Regression model:

[0157]

[0158] in The equipment load-evaporation sensitivity coefficient is derived by fitting historical data. Spare parts safety stock. The calculation introduces climate uncertainty factors. This factor is dynamically adjusted based on the blind confidence interval of the 15-day weather forecast:

[0159]

[0160] When a prolonged period of high temperature and low humidity is predicted (evaporation rate increases by more than 20%), the system automatically increases the safety stock level of vulnerable parts (such as seals and bearings) and generates a purchase recommendation.

[0161] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A predictive maintenance system for salt field production equipment based on artificial intelligence, characterized in that, include: The real-time equipment monitoring and dynamic maintenance cycle modeling module acquires real-time operating data of salt field production equipment and establishes a dynamic maintenance cycle model. The equipment abnormal state trend prediction module predicts the development trend of abnormal states based on the dynamic maintenance cycle model and the historical fault characteristics of the equipment. The critical equipment health status comprehensive assessment and priority ranking module comprehensively assesses and prioritizes the health status of critical equipment based on multi-source sensor data. The intelligent optimization and control module for spare parts inventory management strategies intelligently optimizes and controls spare parts inventory management strategies based on the equipment degradation trends, climate conditions, and changes in evaporation rates output by the predictive model.

2. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 1, characterized in that, The step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes: Collect equipment vibration spectrum data, oil contamination index, current fluctuation parameters, and temperature change curves; Calculate the operational stability index of the equipment ,in The standard deviation of the measurement data is represented by the standard deviation of the measurement data. This represents the mean; By combining historical maintenance records with current operating data, a dynamic maintenance cycle adjustment coefficient is generated. ,in It is the equipment load factor. It's about operational efficiency; The adjustment factor K is used to determine whether to initiate a new maintenance cycle prediction model for optimization.

3. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 1, characterized in that, The prediction of abnormal state development trends based on dynamic maintenance cycle models and historical equipment fault characteristics further includes: Collect the frequency of failures in historical failure modes Mean Time Between Failures and equipment degradation rate ; Calculate the probability of anomalies ,in This is the cumulative time since the last maintenance; A time-series trend prediction network is constructed using an LSTM model to perform multi-step prediction of abnormal development paths; By setting a threshold Determine whether an early warning has been triggered. A high-risk warning is given; otherwise, a low-risk prediction is made. The comprehensive assessment and prioritization of the health status of key equipment based on multi-source sensor data further includes: Extract data from pressure sensors, flow meters, vibration detectors, and temperature probes; Calculate the health index of each device (weight) Signal fluctuation value ); where weight It is the preset equipment importance coefficient; According to health index Rank the equipment and schedule maintenance according to the ranking order; Clustering algorithms are used to divide devices into normal, early warning, and emergency groups for subsequent resource allocation.

4. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 3, characterized in that, The intelligent optimization and control of spare parts inventory management strategies based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes: Extract the degradation rate predicted by the model And the actual monitored evaporation rate of salt fields ; Calculate demand volatility ,in This represents the cumulative historical values ​​over the past 7 days; An improved Markov model is used to simulate inventory demand, and the spare parts turnover rate is defined. ; When spare parts turnover rate Greater than or equal to the threshold A dynamic adjustment mechanism for vehicle inventory.

5. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 4, characterized in that, The step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes: Load rate of collection equipment: Load = (Operating energy consumption / Rated power) ; Introducing seasonal adjustment coefficient ,in This is the average temperature over the past week. For reference temperature; Adjusting maintenance cycles using a normalization formula: CT ,in This is the basic maintenance cycle. This is an empirical coefficient; Set the minimum maintenance cycle Tmin=7 days to prevent over-maintenance.

6. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 5, characterized in that, The prediction of abnormal state development trends based on dynamic maintenance cycle models and historical equipment fault characteristics further includes: Construct a fuzzy inference system, with input variables including device operating time. Number of abnormal features Maintenance interval ; The output variable is the device failure rate. ; The prediction strength of different levels of faults is defined using a fuzzy rule set; Based on expert experience, the prediction results are mapped to tiered alarms (red / yellow / green) to guide response strategies.

7. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 6, characterized in that, The comprehensive assessment and prioritization of the health status of key equipment based on multi-source sensor data further includes: Principal component analysis (PCA) is used to extract common feature components from each sensor data. Define the health weight matrix ,in It is the sensor sensitivity coefficient. It is the reading intensity. It is the exponential coefficient; Based on total weight Classify the equipment; Set maintenance priority criteria: If If it is a high level, then it is a high level. That would be a low level.

8. The predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 7, characterized in that, The intelligent optimization and control of spare parts inventory management strategies based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes: Simulate future weather using historical climate models and estimate potential impacts; Introducing the correlation coefficient of evaporation rate ; Calculate the safety stock of spare parts ,in It is the spare parts shortage tolerance factor. It predicts the frequency of failures; Set upper and lower limits for inventory control. When a supplementary instruction is issued, Adjust inventory strategies as needed.

9. A predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 8, characterized in that, The step of acquiring real-time operating data of salt field production equipment and establishing a dynamic maintenance cycle model further includes: Analyze the vibration frequency domain characteristics of the equipment and identify the resonant frequency band. ; Calculating spectral energy density using a spectrogram Within the range; Calculate the operational stability index of the equipment ; Combined with equipment aging coefficient Establish a dynamic maintenance cycle model ,in This is the basic maintenance cycle.

10. A predictive maintenance system for salt field production equipment based on artificial intelligence according to claim 9, characterized in that, The intelligent optimization and control of spare parts inventory management strategies based on equipment degradation trends, climate conditions, and evaporation rate changes output by the prediction model further includes: Multimodal fusion learning of device image features and vibration data is performed using a convolutional neural network (CNN); Reinforcement learning models are trained based on historical inventory and scheduling data to optimize replenishment strategies; The optimal strategy is generated using the Q-learning algorithm, and a reward function is defined. ; The model is updated based on the reward value after each decision to achieve adaptive inventory management.

Images