An adaptive interlock control method for an ammonia water pump

Abstract

The application relates to the technical field of control, and discloses a self-adaptive interlocking control method for an ammonia water pump, which comprises the following steps: performing multi-parameter real-time monitoring on the ammonia water pump to obtain real-time operation parameters of a main ammonia water pump and a standby pump; using a fault diagnosis model to perform fault screening based on the real-time operation parameters of the main ammonia water pump; performing interlocking control on the main ammonia water pump based on the fault screening result, starting the standby pump within a preset time if it is confirmed that the main ammonia water pump has a fault, opening an outlet valve of the standby pump, cutting off the power supply of the main ammonia water pump, and adjusting the output frequency of a frequency converter to make the outlet pressure smoothly transit; and performing ammonia water pump operation state indication, displaying the start-stop state of the ammonia water pump, automatically displaying a fault code, a recommended measure and a real-time trend diagram when it is confirmed that the main ammonia water pump has a fault. Thus, the safe operation of the ammonia water pump is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of control, in particular to a self-adaptive interlocking control method for ammonia water pump. BACKGROUND

[0002] Synthetic ammonia refers to ammonia synthesized directly from nitrogen and hydrogen under high temperature and pressure and in the presence of a catalyst, which is a basic inorganic chemical process. In modern chemical industry, ammonia is the main raw material for fertilizer industry and basic organic chemical industry, and the demand for ammonia is increasing with the development of science and technology and the needs of agricultural production. In the production of synthetic ammonia, the ammonia recovery tower plays a key role in the recovery of ammonia in the vent gas, and the stable operation of the ammonia water pump, as the core equipment in the ammonia recovery system, is crucial. The ammonia water pump undertakes a key conveying task in the chemical, metallurgical and environmental protection industries, and its operation safety and reliability directly affect the continuity and system safety of the production process. However, the existing ammonia water pump control system still has the following problems in actual application:

[0003] Firstly, the existing ammonia water pump control method lags behind in fault judgment, the pump switching is not smooth, and the operation personnel have difficulty in obtaining accurate decision information in time, resulting in low operation safety and poor stability of the ammonia water pump;

[0004] Secondly, the existing ammonia water pump control system lacks systematic management and state monitoring, which easily leads to maintenance lag, insufficient fault warning and uneven equipment operation;

[0005] Thirdly, in the existing ammonia water pump control, it is easily affected by unauthorized operation or abnormal behavior, which may lead to equipment operation and data management risks, low safety and insufficient controllability. SUMMARY

[0006] The summary section is intended to introduce the concepts in a simplified form, which will be described in detail in the specific embodiments section. The summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0007] The present application proposes a self-adaptive interlocking control method for ammonia water pump to solve one or more of the technical problems mentioned in the background section.

[0008] The present application provides a self-adaptive interlocking control method for ammonia water pump, comprising: using a radar liquid level meter, an acceleration sensor, a pressure transmitter and a temperature sensor to perform multi-parameter real-time monitoring on a plurality of ammonia water pumps, wherein the plurality of ammonia water pumps include a main ammonia water pump and a standby pump, to obtain real-time operation parameters of the plurality of ammonia water pumps;

[0009] The fault diagnosis model is used to perform fault screening based on real-time operation parameters of the main ammonia water pump, and a fault screening result is obtained.

[0010] Interlock control is performed on the main ammonia water pump based on the fault screening result. If it is confirmed that the main ammonia water pump has a fault, the standby pump is started within a preset time, and the output frequency of the frequency converter at the outlet of the ammonia water pump is adjusted to smoothly transition the outlet pressure.

[0011] The start-stop state of the ammonia water pump is displayed on the operation interface of the DCS system. When it is confirmed that the main ammonia water pump has a fault, the fault code, recommended measures, and a real-time trend graph are displayed.

[0012] Optionally, a radar liquid level meter, an acceleration sensor, a pressure transmitter, and a temperature sensor are used to perform multi-parameter real-time monitoring on multiple ammonia water pumps, wherein the multiple ammonia water pumps include the main ammonia water pump and the standby pump, to obtain real-time operation parameters of the multiple ammonia water pumps, including:

[0013] The radar liquid level meter is used to monitor the inlet flow of the main ammonia water pump and the standby pump in real time.

[0014] The acceleration sensor is used to collect pump body vibration data of the main ammonia water pump and the standby pump. The mechanical faults of the main ammonia water pump and the standby pump are identified by analyzing the pump body vibration data.

[0015] The temperature sensor is used to monitor the bearing temperature and the motor stator temperature of the main ammonia water pump and the standby pump, and temperature data is collected.

[0016] The pressure transmitter is used to detect the inlet and outlet pressures of the main ammonia water pump and the standby pump in real time. Abnormal pressure fluctuations exceeding a preset pressure threshold are used as a fault judgment basis.

[0017] Optionally, a fault diagnosis model is used to perform fault screening based on real-time operation parameters of the main ammonia water pump, including:

[0018] The historical operation parameters of the main ammonia water pump are obtained. Time domain analysis and frequency domain analysis are used to extract signal features, and feature dimensionality reduction is performed on the historical operation parameters. The historical operation parameters are divided into transient parameters and steady-state parameters.

[0019] The historical fault data of the main ammonia water pump are obtained. Migration learning optimization is performed based on the historical fault data. The weight distribution of key features is strengthened in combination with an attention mechanism, and a machine learning model is constructed.

[0020] A sliding window statistical threshold is used for the steady-state parameters, and a delay confirmation mechanism is introduced for the transient parameters. The adjusted machine learning model is confirmed as a fault diagnosis model by fusing multi-parameter abnormal confidence through Bayesian inference.

[0021] Based on the real-time operation parameters of the main ammonia water pump, a fault diagnosis model is used to perform fault diagnosis.

[0022] Determine the abnormal parameters in the real-time operating parameters of the main ammonia water pump, analyze the causal relationship between the abnormal parameters, lock the fault root cause, and generate a hierarchical alarm signal according to the fault influence range and emergency level.

[0023] Optionally, based on the fault screening result, the main ammonia water pump is interlocked controlled, if it is confirmed that the main ammonia water pump has a fault, the standby pump is started within a preset time, the output frequency of the frequency converter at the outlet of the ammonia water pump is adjusted to smoothly transition the outlet pressure, including:

[0024] If it is confirmed through the fault diagnosis model that the main ammonia water pump has an unrecoverable fault, the fault signal waveform is continuously tracked within a preset time window, the synchronicity and trend correlation of multi-channel parameter changes are analyzed, the signal attenuation characteristics and environmental interference mode difference are verified, the redundant sensor data is cross-compared and matched with the historical interference case library, the transient abnormal pulse or random noise interference is filtered, and the interlock protection mechanism is triggered after confirming the fault persistence;

[0025] After confirming no misjudgment, a start instruction is sent to the control system of the standby pump, and the standby pump preloads the operating parameters before the main ammonia water pump stops;

[0026] The power supply of the driving motor of the main ammonia water pump is cut off, the driving motor of the standby pump is soft-started according to a preset acceleration curve, and the outlet valve is controlled to slowly open to the fully open position;

[0027] The current running ammonia water pump state is updated to the standby pump.

[0028] Optionally, the start-stop state of the ammonia water pump is displayed on the operation interface of the DCS system, when it is confirmed that the main ammonia water pump has a fault, the fault code, the recommended measures and the real-time trend graph are displayed, including:

[0029] A dynamic indication module of the ammonia water pump is embedded on the operation interface of the DCS system, for displaying the real-time operating parameters of the main ammonia water pump and the standby pump;

[0030] According to the fault code, the hierarchical alarm signal and the preset knowledge base, the recommended measures are determined and displayed on the operation interface of the DCS system;

[0031] A hierarchical window is popped up on the operation interface of the DCS system, the fault code and the text description are displayed on the left side, the real-time trend graph is embedded on the right side, the historical data curve of the same working condition is superimposed to assist in judging the fault evolution trend, and the disposal step buttons with priority sorting are displayed at the bottom;

[0032] When it is determined that the main ammonia water pump has a recoverable fault, a preset first-level alarm function is triggered, triggering a yellow rotating light and a low-frequency beep, and when it is determined that the main ammonia water pump has an unrecoverable fault, a second-level alarm function is started, triggering a red flashing light and an alarm sound.

[0033] Optionally, an adaptive interlocking control method for an ammonia pump further includes:

[0034] Record the fault trigger time, number of switching times, and changes in operating conditions before and after switching, and store them in a relational database after being timestamped;

[0035] Based on association rules, perform correlation analysis on relational databases and output a fault troubleshooting list;

[0036] Based on the fault diagnosis list, corresponding work orders are generated, tasks are automatically assigned to designated maintenance teams, and spare parts requisition lists are pushed to the warehouse system.

[0037] The maintenance process data and original fault records are uploaded in real time via mobile terminals to generate a maintenance effectiveness evaluation index.

[0038] Automatically generate electronic work orders containing timestamps, fault codes, and handling operations, store them in the database, and add the fault characteristics to the machine learning training set.

[0039] Optionally, the adaptive interlocking control method for an ammonia pump of the present invention further includes:

[0040] Based on the real-time operating parameters of the main ammonia pump and the standby pump, determine the health status level of the main ammonia pump and the standby pump;

[0041] The health status level is compared with the historical health status level to generate a comparison result; if the comparison result indicates that the ammonia pump has been downgraded, an early warning is generated and a planned maintenance is arranged.

[0042] Optionally, the adaptive interlocking control method for an ammonia pump of the present invention further includes:

[0043] Record the cumulative operating time of the main ammonia pump and the standby pump separately. The cumulative operating time includes the main operating time and the standby time.

[0044] When the main operating time reaches the preset threshold and the main operating time is greater than or equal to the standby time, the main ammonia pump and the standby pump will be rotated.

[0045] During the rotation process, the same pressure smooth transition control method as that used for fault switching is adopted;

[0046] After the rotation is completed, the status of the main ammonia pump will be updated to standby pump.

[0047] The present invention has the following beneficial effects:

[0048] 1. Improved safety and stability of ammonia pump operation. Specifically, through multi-parameter monitoring and fault diagnosis models, recoverable and unrecoverable faults can be distinguished, reducing false positives and false negatives. Automatic interlocking control combined with frequency converter adjustment of outlet pressure allows the standby pump to seamlessly take over in the event of a main pump failure, avoiding drastic fluctuations in process pressure. The DCS system displays pump status, fault codes, trend graphs, and suggested measures, enabling operators to promptly assess fault evolution and implement maintenance measures. This reduces reliance on manual operation, lowers the risk of equipment damage, and ensures continuous and stable operation of the ammonia pump system.

[0049] 2. Improved maintenance efficiency and response speed. Specifically, through monitoring and analysis of faults, health status, and uptime, closed-loop management of pump unit status is achieved: automatically detecting abnormal or degraded pump units, generating work orders or early warning prompts, and executing rotation and pressure smoothing control to ensure reliable operation, timely maintenance, efficient allocation, and balanced lifespan of pump units, while supporting data accumulation for model optimization and subsequent decision-making.

[0050] 3. Improved data transmission security and controllability. Specifically, through multiple layers of security measures such as data encryption, firewall filtering, operational behavior auditing, pattern analysis, isolation gateways, and access authentication, an end-to-end industrial control system security protection and abnormal operation detection system is formed, thereby ensuring data integrity, operational traceability, and access legitimacy. Attached Figure Description

[0051] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0052] Figure 1 This is a flowchart of an adaptive interlocking control method for an ammonia pump according to the present invention;

[0053] Figure 2 This is a schematic diagram of the ammonia pump control interface of an adaptive interlocking control method for an ammonia pump according to the present invention.

[0054] Figure 3 This is another schematic diagram of an adaptive interlocking control method for an ammonia pump according to the present invention;

[0055] Figure 4 This is another schematic diagram of an adaptive interlocking control method for an ammonia pump according to the present invention. Detailed Implementation

[0056] The invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the invention. It should be understood that the drawings and embodiments of the invention are for illustrative purposes only and are not intended to limit the scope of protection of the invention.

[0057] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0058] It should be noted that the concepts of "first" and "second" mentioned in this invention are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0059] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0060] The names of messages or information exchanged between the various devices of this invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0061] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0062] like Figure 1 The diagram shows a flowchart of an adaptive interlocking control method for an ammonia pump according to the present invention, which specifically includes the following steps:

[0063] Step S101: Use radar level gauge, accelerometer, pressure transmitter and temperature sensor to monitor multiple ammonia pumps in real time with multiple parameters. The multiple ammonia pumps include main ammonia pump and standby pump to obtain the real-time operating parameters of multiple ammonia pumps.

[0064] In some embodiments, the execution entity of the adaptive interlocking control method for an ammonia pump according to the present invention is a backend server. A radar level gauge is a sensing device that measures liquid level using the principle of electromagnetic wave reflection. Its advantages include non-contact measurement, high accuracy, and strong anti-interference capability. An accelerometer is a sensing device that detects the acceleration of vibration or motion of an object. In ammonia pump applications, it is usually installed on the pump body or bearing housing to collect vibration signals. By performing spectral analysis on the collected signals, potential mechanical faults such as rotor imbalance, bearing wear, and cavitation can be identified. A pressure transmitter is a sensor device that converts medium pressure into a standard electrical signal output. The pressure transmitter is used to detect the inlet and outlet pressures of the ammonia pump in real time and transmit the pressure signal to the backend server. When the detected value exceeds a set threshold, it can be used as a basis for anomaly judgment. Temperature sensors, including thermocouples, resistance thermometers, or infrared thermometers, are used to monitor the temperature parameters of key parts of the pump body in real time. The main ammonia pump refers to the working pump that undertakes the task of ammonia transportation during normal operation. The standby pump starts when the main pump fails, is under maintenance, or is under repair to ensure the continuity and stability of the ammonia water delivery process. Based on this, a radar level gauge is used to monitor the inlet flow rate of both the main and standby ammonia water pumps in real time; an accelerometer collects vibration data of the pump bodies of both pumps, and analysis of this data identifies mechanical faults in the main and standby pumps; temperature sensors monitor the bearing and motor stator temperatures of both pumps, and a pressure transmitter is used to monitor the inlet and outlet pressures of both pumps in real time, using abnormal pressure fluctuations exceeding preset pressure thresholds as a basis for fault diagnosis. Real-time operating parameters include liquid level (or flow rate), pump vibration data, inlet and outlet pressures, bearing temperature, and motor temperature, which serve as input data for subsequent fault diagnosis models.

[0065] Step S102: Use the fault diagnosis model to screen faults based on the real-time operating parameters of the main ammonia pump and obtain the fault screening results.

[0066] In some embodiments, historical operating parameters of the main ammonia pump are acquired, and signal features are extracted using time-domain and frequency-domain analysis. The historical operating parameters are then dimensionality-reduced and divided into transient and steady-state parameters. Historical fault data of the main ammonia pump is acquired, and transfer learning optimization is performed based on this data. An attention mechanism is used to strengthen the weight allocation of key features, and a machine learning model is constructed. A sliding window statistical threshold is used for steady-state parameters, and a delayed confirmation mechanism is introduced for transient parameters. Bayesian inference is used to fuse multi-parameter anomaly confidence levels, and the adjusted machine learning model is confirmed as a fault diagnosis model. Based on the real-time operating parameters of the main ammonia pump, the fault diagnosis model is used for fault diagnosis. Abnormal parameters in the real-time operating parameters of the main ammonia pump are identified, the causal relationships between abnormal parameters are analyzed, the root cause of the fault is located, and graded alarm signals are generated according to the scope and urgency of the fault. The fault diagnosis model refers to a machine learning model trained based on historical operating data, historical fault cases, and feature parameters of the ammonia pump. It is used to analyze the collected multiple parameters during real-time operation to determine whether anomalies exist and their nature. The model can be a deep learning model based on artificial intelligence, such as a convolutional neural network (CNN), a long short-term memory network (LSTM), or a time-series model based on an attention mechanism. The core function of the model is to extract and compare features from the input multi-parameter data (such as pressure, vibration, temperature, etc.) and output a judgment result. Fault screening refers to comparing and calculating real-time operating parameters using a fault diagnosis model to screen for potential fault signals and filter out false alarms caused by environmental noise, transient disturbances, etc. Fault screening results can be: the main ammonia pump is operating normally; the main ammonia pump has a recoverable fault (e.g., slight vibration exceeding limits, short-term temperature rise); the main ammonia pump has an unrecoverable fault (e.g., severe bearing wear, outlet pressure failure) and the corresponding confidence level.

[0067] Step S103: Based on the fault screening results, the main ammonia pump is interlocked and controlled. If the main ammonia pump is confirmed to be faulty, the standby pump is started within a preset time and the frequency of the inverter at the outlet of the ammonia pump is adjusted to make the outlet pressure transition smoothly.

[0068] In some embodiments, if an unrecoverable fault is confirmed in the main ammonia pump through a fault diagnosis model, the fault signal waveform is continuously tracked within a preset time window. The synchronicity and trend correlation of multi-channel parameter changes are analyzed, and the differences in signal attenuation characteristics and environmental interference patterns are verified. Data from redundant sensors is cross-referenced and matched against a historical interference case library to filter transient abnormal pulses or random noise interference. After confirming the persistence of the fault, an interlocking protection mechanism is triggered. If no misjudgment is confirmed, a start command is sent to the control system of the standby pump, and the standby pump preloads the operating parameters of the main ammonia pump before shutdown. The power supply to the drive motor of the main ammonia pump is cut off; the drive motor of the standby pump soft-starts according to a preset acceleration curve and controls the outlet valve to slowly open to the fully open position. The status of the currently operating ammonia pump is updated to standby pump. Here, interlocking control refers to an automated control mechanism used to ensure safe operation when the main pump fails, avoiding pressure fluctuations or equipment damage. The preset time refers to the time window from confirming the main pump fault to starting the standby pump, which can be set according to process requirements (e.g., 1 second to 5 seconds or longer), used to verify the persistence of the fault and avoid transient misjudgments. The output frequency of a frequency converter refers to the speed at which the frequency converter controls the pump motor, thereby changing the pump outlet flow rate and pressure by adjusting the frequency. Smooth outlet pressure transition refers to the process of switching from the main pump to the standby pump, where the outlet pressure is smoothly transitioned from the main pump operating condition to the standby pump operating condition by controlling the pump speed and the operation of the outlet valve, avoiding pressure fluctuations that could impact the pipeline and process.

[0069] Step S104: Display the start / stop status of the ammonia pump on the DCS system's operation interface. When a fault is confirmed in the main ammonia pump, display the fault code, suggested measures, and real-time trend graph.

[0070] In some embodiments, an ammonia pump dynamic indicator module is embedded in the DCS system's operating interface to display real-time operating parameters of the main ammonia pump and the standby pump. Based on fault codes, graded alarm signals, and a preset knowledge base, recommended measures are determined and displayed on the DCS system's operating interface. A layered window pops up on the DCS system's operating interface, displaying the fault code and text description on the left, embedding a real-time trend graph on the right, overlaying historical data curves under the same operating conditions to assist in judging the fault evolution trend, and displaying priority-ordered handling step buttons at the bottom. When a recoverable fault is determined in the main ammonia pump, a preset level-one alarm function is triggered, activating a yellow rotating light and a low-frequency buzzer. When an unrecoverable fault is determined in the main ammonia pump, a level-two alarm function is activated, triggering a red flashing light and an alarm sound. The DCS system (Distributed Control System) is used for real-time monitoring and control of industrial field equipment. Its functions include: data acquisition, status display, alarm prompts, and operational interaction. The DCS is a display and interaction platform that receives data from the backend server and displays it to the operator. The ammonia pump start / stop status refers to the current operating status of the main pump and the standby pump, such as "running," "stopping," or "starting." Fault codes, generated by the backend server, identify specific fault types or levels (e.g., recoverable / unrecoverable faults) of the main ammonia pump. This facilitates unified system management and rapid fault identification. The backend server monitors the pump motor status and sends the information to the DCS (Distributed Control System) via the network. Recommended actions are operational suggestions or maintenance plans generated based on the fault codes and a preset knowledge base, such as "check pump supports" or "stop and check the cooling system." These are generated by the backend server and sent to the DCS for display. Real-time trend graphs display curves of key parameters changing over time, such as pump outlet pressure, pump vibration, and temperature changes, helping operators determine the evolution trend of the fault and take appropriate measures.

[0071] Among them, such as Figure 3 As shown, multiple ammonia pumps are monitored in real time using radar level gauges, accelerometers, pressure transmitters, and temperature sensors. These pumps include main ammonia pumps and standby pumps. The real-time operating parameters of these pumps are obtained, including:

[0072] Step S201: Use a radar level gauge to monitor the inlet flow rate of the main ammonia pump and the standby pump in real time;

[0073] In some embodiments, a radar level gauge is installed at the top of the reservoir or pipe at the pump inlet to measure the liquid level. The inlet flow rate is calculated using a physical formula, taking into account the cross-sectional area of ​​the pump inlet pipe and the rate of change of liquid level.

[0074] ;

[0075] in, For inbound traffic, This is the cross-sectional area of ​​the pump inlet pipe. The rate of change of liquid level over time is used to calculate the inlet flow rate data, which is then transmitted to the backend server for monitoring and analysis.

[0076] Step S202: Collect pump body vibration data of the main ammonia pump and the standby pump through an accelerometer, and identify mechanical faults of the main ammonia pump and the standby pump by analyzing the pump body vibration data.

[0077] In some embodiments, the accelerometer can be a capacitive, piezoelectric, or MEMS miniature accelerometer. Accelerometers are installed in key parts of the main ammonia pump and standby pump (such as bearing housings and pump casings) to collect vibration signals in real time. The collected vibration signals are transmitted to a backend server via an A / D converter. The backend server performs a Fourier transform on the vibration signals to extract frequency domain features: bearing faults typically manifest as high-frequency characteristic peaks, while impeller imbalance manifests as an increase in fundamental frequency amplitude. Pump body vibration data refers to the mechanical vibration information generated by the pump during operation, including parameters such as amplitude (intensity), frequency (periodicity), and phase. The vibration data reflects the mechanical state of components such as pump bearings, impellers, and bushings. Mechanical faults in the main ammonia pump and standby pump are determined by extracting the frequency domain features. Mechanical faults refer to abnormalities in the pump's moving parts, such as bearing wear or damage, impeller imbalance, etc.

[0078] Step S203: Monitor the bearing temperature and motor stator temperature of the main ammonia pump and the standby pump using temperature sensors to collect temperature data.

[0079] In some embodiments, thermocouples or resistance thermometers are installed on the bearing housings or stator surfaces of the main ammonia pump and the standby pump. Temperature signals are acquired in real time and transmitted to a backend server via an industrial network (such as Modbus or PROFIBUS). Bearing temperature refers to the temperature of the pump bearing portion; excessively high bearing temperatures may indicate insufficient lubrication, wear, or malfunction. Motor stator temperature refers to the temperature of the stator windings of the pump drive motor; excessively high temperatures may indicate overload, poor motor heat dissipation, or insulation damage.

[0080] Step S204: Use a pressure transmitter to monitor the inlet and outlet pressures of the main ammonia pump and the standby pump in real time, and use abnormal pressure fluctuations that exceed the preset pressure threshold as the basis for fault judgment.

[0081] In some embodiments, pressure transmitters are installed on the pump inlet and outlet pipelines. The pressure transmitters transmit pressure values ​​to a backend server in real time. The backend server presets pressure thresholds; for example, an inlet pressure below 0.2 MPa indicates that the main ammonia pump and standby pump may be cavitating; an outlet pressure above 1.5 MPa indicates that the main ammonia pump and standby pump may be blocked at the outlet. When the pressure value exceeds the threshold, it is marked as a potential fault. Inlet and outlet pressures include inlet and outlet pressures. Inlet pressure refers to the pressure at the pump inlet, used to determine if the pump is cavitating or has abnormal suction. Outlet pressure refers to the pressure at the pump outlet, used to determine if the pump's delivery capacity is normal and if there is any blockage or leakage. Abnormal pressure fluctuations refer to pressure changes exceeding preset safety thresholds or abnormal fluctuations, such as rapid drops, pulsations, or sudden increases. This can reflect pump idling, blockage, valve malfunction, or system abnormalities. The preset pressure thresholds are upper and lower pressure limits set in advance based on the ammonia pump design parameters, system pipeline characteristics, and safe operation requirements. These thresholds are used to determine if the pump operation is abnormal; exceeding the threshold triggers a fault or alarm.

[0082] Among them, such as Figure 4 As shown, a fault diagnosis model is used to screen for faults based on the real-time operating parameters of the main ammonia pump, including:

[0083] Step S301: Obtain the historical operating parameters of the main ammonia pump, extract signal features using time domain analysis and frequency domain analysis, perform feature dimensionality reduction on the historical operating parameters, and divide the historical operating parameters into transient parameters and steady-state parameters;

[0084] In some embodiments, historical operating parameters refer to various monitoring data of the main ammonia pump during its past operation, including: flow rate (inlet flow rate, outlet flow rate), pressure (inlet and outlet pressure), temperature (bearing temperature, motor stator temperature), and pump body vibration data. These data are typically collected in real-time by sensors and stored in a database on a backend server. Based on this, the historical operating parameters of the main pump are retrieved from the database. For each parameter in the historical operating parameters, the calculated mean, RMS, peak value, kurtosis, and other indicators are extracted to determine time features. An FFT is performed on each parameter in the historical operating parameters to extract the main frequency components, harmonics, and frequency amplitudes. The feature matrix is ​​then dimensionality-reduced (e.g., by principal component analysis), retaining the main components for subsequent fault diagnosis model training. Based on the data time period and operating status, the features are divided into transient parameters and steady-state parameters; for example, during the start-up / shutdown phase, they are transient parameters; during continuous operation, they are steady-state parameters. Time-domain analysis refers to analyzing the changes of a signal over time to extract statistical features such as mean, root mean square (RMS), peak value, kurtosis, and skewness. This can reveal long-term trends, sudden pulses, or changes in vibration intensity during pump operation. Frequency-domain analysis involves converting the signal from the time domain to the frequency domain (e.g., through Fourier transform (FFT)) and analyzing its frequency components. This can identify periodic faults such as bearing failures or impeller imbalances. Feature reduction involves compressing a large number of correlated or redundant features from the original historical data into fewer key features to improve model training efficiency and accuracy. This can include principal component analysis, linear discriminant analysis, or autoencoders. Transient parameters refer to short-term fluctuations during pump startup, shutdown, or load changes, reflecting instantaneous anomalies or shocks. Steady-state parameters refer to the characteristics of the pump during continuous stable operation, used to determine long-term trends and stable performance.

[0085] Step S302: Obtain historical fault data of the main ammonia pump, perform transfer learning optimization based on the historical fault data, strengthen the weight allocation of key features by combining the attention mechanism, and build a machine learning model.

[0086] In some embodiments, historical fault data refers to records of past faults in the main ammonia pump, including fault type (e.g., bearing damage, impeller imbalance, seal leakage), fault occurrence time, and pre-fault operating parameters (flow rate, pressure, temperature, vibration, etc., fault impact range and severity). This data is pre-stored on a backend server. Based on this, the historical fault data is organized into an input feature vector (including vibration, pressure, temperature, and flow rate features). The parameters of an existing pump type or a similar pump fault diagnosis model are used as initial weights (transfer learning). The parameters of an existing pump type or a similar pump fault diagnosis model refer to the optimized weights and bias parameters in a pre-trained machine learning or deep learning model; these parameters reflect the mapping relationship between pump operating status and faults. Based on this, a deep neural network model is constructed: Input layer: features from various sensors; Intermediate layer: multi-layer fully connected or convolutional / recurrent units; Attention layer: dynamically adjusts feature weights to highlight key features; Output layer: Fault type or fault probability. The model is trained using historical fault data, and parameters are optimized to enable the model to accurately predict new fault types or states. Transfer learning refers to transferring knowledge learned by an existing model from one dataset or task to another related task, improving model training efficiency or accuracy. In this application: fault model parameters of similar pump types or pumps under the same operating conditions can be used as initial weights to avoid training the model from scratch and improve diagnostic accuracy with small sample fault data. Attention mechanisms are a technique in machine learning used to dynamically adjust the importance weights of different input features. In fault diagnosis, certain sensor features (such as vibration peaks) are more important for identifying bearing faults; attention mechanisms can strengthen the weights of these key features in model calculations, improving diagnostic accuracy. Machine learning models refer to models that learn patterns through training data and are used for real-time fault diagnosis. Deep neural networks, convolutional neural networks (processing time-series or frequency-domain features), and recurrent neural networks (capturing time-series trends) can be used.

[0087] Step S303: For steady-state parameters, a sliding window statistical threshold is used, and a delayed confirmation mechanism is introduced for transient parameters. By fusing multi-parameter anomaly confidence through Bayesian inference, the adjusted machine learning model is confirmed as a fault diagnosis model.

[0088] In some embodiments, the steady-state parameters can be average flow rate, average inlet and outlet pressure, and average motor stator temperature. Transient parameters include, for example, instantaneous peak values ​​of pump body vibration, instantaneous flow rate fluctuations, and instantaneous pressure pulses. Based on this, the steady-state parameters are statistically analyzed by time segments (windows), such as calculating the mean and standard deviation. For transient parameters, a time delay is set for confirmation; if a transient anomaly persists for more than the set time, it is considered a fault. The probability of each parameter's anomaly is input into a Bayesian fusion calculation to determine the overall fault confidence. The model trained and adjusted using sliding window, time delay confirmation, and Bayesian fusion is determined as the fault diagnosis model. Here, the anomaly confidence refers to the probabilistic quantitative index of the system's judgment of a certain operating parameter or overall result as "abnormal / faulty," typically ranging from 0 to 1. As an example, for a steady-state parameter with an average pressure of 1.0 MPa, if the sliding window statistics show that the average pressure over the most recent minute is 1.12 MPa, which is greater than the threshold of 1.1 MPa, then an anomaly is determined. Transient parameters: If the vibration peak instantaneously reaches 3 mm / s but only lasts for 1 second, it is not considered abnormal; if it lasts for 6 seconds and exceeds the threshold, it is considered abnormal. Fusion results: The probability of pressure abnormality is 0.7, the probability of vibration abnormality is 0.85, and the overall fault confidence after Bayesian fusion is 0.82, so it is confirmed as a fault.

[0089] Step S304: Based on the real-time operating parameters of the main ammonia pump, use a fault diagnosis model to perform fault diagnosis.

[0090] Step S305: Determine the abnormal parameters in the real-time operating parameters of the main ammonia pump, analyze the causal relationship between the abnormal parameters, pinpoint the root cause of the fault, and generate graded alarm signals according to the scope of the fault's impact and its urgency.

[0091] In some embodiments, real-time acquired multi-parameter data is input into the fault diagnosis model. The model extracts features from the input signals (e.g., time-domain and frequency-domain features) and compares them with normal / fault feature patterns obtained during training. The model output includes: normal, recoverable fault, unrecoverable fault, and corresponding anomaly confidence. Based on the model output, abnormal parameters are determined. Causal reasoning (e.g., Bayesian networks, time series correlation analysis) is used to identify root cause faults. Severity is measured based on the magnitude and duration of abnormal parameters deviating from thresholds. Based on multi-parameter causal relationships and the pump unit's location in the system, it is analyzed whether the anomaly affects only a single pump or extends to the delivery pipeline, ammonia storage tank, or even upstream / downstream equipment. Combining the urgency level and the scope of the fault's impact, a pre-stored graded alarm table is consulted to determine the alarm level. The graded alarm table includes severity, scope of impact, and corresponding alarm level and alarm method. Abnormal parameters refer to operating parameters that deviate from the normal threshold range, such as overpressure, overtemperature, and over-vibration. Causal relationship analysis determines which anomaly is the "root cause" and which are "accompanying phenomena" by analyzing the sequence and logical relationship between abnormal parameters. Tiered alarm signals: Different levels of alarms are output according to the urgency of the fault, such as Level 1 warning (recoverable) and Level 2 alarm (non-recoverable). The root cause of the fault refers to the direct or primary cause of the abnormal operation of the ammonia pump, which is different from accompanying abnormalities (secondary phenomena). It is the source of the fault factors that cause a series of apparent abnormalities in the system, such as vibration, pressure fluctuations, and temperature rise.

[0092] The system includes interlocking control of the main ammonia pump based on fault screening results. If a fault is confirmed in the main ammonia pump, the standby pump is started within a preset time. The frequency converter output of the ammonia pump outlet is adjusted to ensure a smooth transition of the outlet pressure, including:

[0093] If the fault diagnosis model confirms that the main ammonia pump has an unrecoverable fault, the fault signal waveform is continuously tracked within a preset time window. The synchronicity and trend correlation of the changes in multi-channel parameters are analyzed, the differences in signal attenuation characteristics and environmental interference modes are verified, and the data from redundant sensors are cross-compared with historical interference case databases to filter transient abnormal pulses or random noise interference. After confirming the persistence of the fault, the interlocking protection mechanism is triggered.

[0094] The interlocking protection mechanism is as follows: after confirming that there is no misjudgment, a start command is sent to the control system of the standby pump, and the standby pump is preloaded with the operating parameters of the main ammonia pump before shutdown.

[0095] Disconnect the power supply to the drive motor of the main ammonia pump; the drive motor of the standby pump will start softly according to the preset acceleration curve, and the outlet valve will be slowly opened to the fully open position.

[0096] Update the status of the currently running ammonia pump to standby pump.

[0097] In some embodiments, an unrecoverable fault refers to a fault in the ammonia pump that, once it occurs, cannot be restored to normal operation through short-term adjustments, manual intervention, or system self-healing mechanisms. This differs from a "recoverable fault": recoverable faults (such as momentary insufficient liquid supply or transient false alarms from sensors) can be recovered through manual or automatic adjustments. Unrecoverable faults require stopping the pump and switching to a standby pump. A preset time window refers to a set time period (e.g., 2s to 10s). Based on this, if an unrecoverable fault is confirmed in the main ammonia pump, a moving average or weighted statistical analysis is performed within the preset time window to filter out single-point anomalies. For example, an "anomaly" is determined only if the pressure is below a threshold for five consecutive sampling periods. First, multi-channel real-time operating parameters of the main ammonia pump, such as vibration signals, inlet and outlet pressures, bearing temperatures, and motor current, are collected within the preset time window, and corresponding signal waveforms are generated. Through time-domain and frequency-domain analysis, the mean, RMS value, main frequency components, and amplitude of the signals are extracted to determine the persistence of the fault signal. Second, the synchronicity and trend correlation of changes in multi-channel parameters are analyzed. For example, if the vibration signal significantly increases while the outlet pressure decreases and the flow rate decreases, a causal relationship can be established, indicating a high probability of equipment failure. Next, the differences between signal attenuation characteristics and environmental interference patterns are verified. If the abnormal signal attenuates rapidly and lacks stability within a short period, it may be electromagnetic interference or random noise and should be excluded. If the abnormal signal continuously increases and conforms to typical mechanical failure patterns, it is determined to be a genuine failure. Furthermore, data from redundant sensors are cross-referenced. If the results of similar parameters (such as outlet pressure) collected by multiple sensors are basically consistent, the reliability of the fault diagnosis is enhanced. If only a single sensor detects an anomaly, and the redundant channels do not show corresponding anomalies, it can be determined to be a sensor error. Simultaneously, the current abnormal pattern is matched with a historical interference case database. If the matching result indicates that the anomaly belongs to a typical interference pattern, the interlocking protection is not triggered. If the matching result is consistent with a known unrecoverable fault mode, the persistence of the fault is confirmed. After confirming the persistence of the fault, the interlock protection mechanism is triggered, including: after confirming there is no misjudgment, cutting off the power supply to the main ammonia pump drive motor, sending a start command to the standby pump control system, and loading the operating parameters of the main pump before shutdown (such as pressure setpoint and frequency) onto the standby pump. The standby pump performs a soft start according to a preset acceleration curve: the inverter accelerates according to the acceleration curve while the outlet valve is slowly opened to avoid pressure surges, thereby achieving a smooth switch from the main pump to the standby pump and ensuring stable system outlet pressure. Finally, the "currently running pump" is marked as the standby pump to ensure the correctness of subsequent control and monitoring logic. The continuous tracking within the preset time window does not determine a fault based on a single abnormal data point, but rather continuously collects and analyzes signals within a set time period to avoid misjudgment.The synchronicity and trend correlation of multi-channel parameter changes are used to determine whether different monitored parameters (pressure, temperature, vibration, flow) change simultaneously and whether the trend conforms to typical fault characteristics. Verifying the difference between signal attenuation characteristics and environmental interference patterns aims to distinguish between "real faults" and "external interference." Signal attenuation characteristics refer to the gradual deterioration of signals in mechanical faults, which do not disappear instantly. Environmental interference patterns, such as electromagnetic interference / instantaneous power grid fluctuations, can cause single-pulse anomalies, but without a sustained trend. Redundant sensor data cross-comparison involves deploying multiple sensors at the same monitoring point or using different measurement methods to collect similar information and comparing them. For example, if two pressure transmitters collect the same outlet pressure, and only one shows an anomaly, it may be a sensor fault. Simultaneous anomalies from vibration sensors and motor current monitoring are more reliable. Historical interference case library matching compares the current abnormal waveform with a pre-stored "interference feature library." For example, if the waveform matches the characteristics of an "electromagnetic interference pulse," it is removed as interference. If the waveform matches the typical pattern of "bearing damage," it is retained as a valid fault. The interlocking protection mechanism is triggered immediately once the fault is confirmed to be real, unrecoverable, and persistent.

[0098] The DCS system's operation interface displays the start / stop status of the ammonia pump. When a fault is confirmed in the main ammonia pump, it displays the fault code, suggested measures, and real-time trend graphs, including:

[0099] A dynamic indicator module for ammonia pumps is embedded in the DCS system's operating interface to display the real-time operating parameters of the main ammonia pump and the standby pump.

[0100] Based on the fault codes, graded alarm signals, and preset knowledge base, determine the recommended measures and display them on the DCS system's operation interface;

[0101] A layered window pops up on the DCS system's operating interface. The left side displays the fault code and text description, the right side embeds a real-time trend chart, and overlays historical data curves under the same working conditions to help judge the fault evolution trend. The bottom displays handling steps buttons with priority sorting.

[0102] When a recoverable fault is detected in the main ammonia pump, the preset first-level alarm function is triggered, which activates the yellow rotating light and low-frequency buzzer. When an unrecoverable fault is detected in the main ammonia pump, the second-level alarm function is activated, which activates the red flashing light and alarm sound.

[0103] In some embodiments, the dynamic indicator module refers to a visualization component on the DCS interface, used to display the operating parameters of the equipment in real time and dynamically update the display status as the parameters change. A dynamic indicator module is created on the DCS system's operating interface. Through a data interface with the backend server, it acquires sensor data from the main pump and standby pump in real time and displays this data within the module in the form of trend curves, digital instruments, status indicator lights, etc. When parameters become abnormal or exceed preset thresholds, the module automatically updates the display status (e.g., color change or flashing) to alert the operator. The backend server, based on the results output by the fault diagnosis model, queries the fault code mapping table to determine the corresponding fault code, and then matches the corresponding fault handling plan in the knowledge base to generate suggested measures. These suggested measures are displayed to the operator through the DCS operating interface in the form of text, buttons, or prompts. For example, if the fault code is E101 and it is a level 2 alarm, the interface will display "Bearing temperature too high, stop immediately and check the lubrication system," and provide "Confirm Stop" or "Switch to Standby Pump" operation buttons. The fault codes identify the specific fault type of the main ammonia pump or standby pump. For example, "E101" indicates a bearing overheating alarm, which is a graded alarm level based on the severity of the fault. A level 1 alarm indicates a recoverable fault, while a level 2 alarm indicates an unrecoverable fault. The preset knowledge base stores fault handling experience and standard operating procedures on the backend server, including handling solutions for different fault codes. Suggested measures are automatically generated operation guidelines or handling solutions based on the fault type and alarm level, guiding operators to take appropriate measures. The DCS operation interface divides the display area into multiple levels or modules, each displaying different types of information, facilitating the simultaneous presentation of fault information, trend data, and operation buttons. The left side displays the fault code and text description: visually informing the operator of the pump's fault type and a brief description, such as "E101: Bearing overheating." The right side embeds a real-time trend graph: displaying the curves of key operating parameters (such as pressure, flow rate, and temperature) of the main and standby pumps over time, updated in real time. Historical data curves under the same operating conditions are overlaid on the trend graph: historical operating data (such as pressure change curves under the same operating conditions in the past) are overlaid to facilitate comparison and judgment of fault evolution trends. The bottom displays priority-ordered action buttons: buttons arranged according to operation priority, such as "Switch to Standby Pump" and "Check Lubrication System," for quick operator execution. If the main ammonia pump has a recoverable fault, a Level 1 alarm is triggered, with a flashing yellow rotating light and a low-frequency buzzer. If the main pump has an unrecoverable fault, a Level 2 alarm is triggered, with a flashing red strobe light and a high-decibel alarm, while the interface prompts "Switch to Standby Pump Immediately." The operator can then perform the corresponding operation according to the provided buttons, thus achieving unified management of fault information visualization, alarm classification, and guided operation. The Level 1 alarm function is a warning method set for recoverable faults, used to alert the operator but not requiring immediate shutdown.The Level 2 alarm function is a warning system designed for irreversible faults, prompting operators to take immediate emergency measures. A yellow rotating light and a low-frequency buzzer correspond to the Level 1 alarm's visual and audible alerts; the color and audio frequency are designed for low-level warnings. A red flashing light and an alarm sound correspond to the Level 2 alarm's high-level visual and audible alerts, used for emergency warnings.

[0104] These embodiments improve the safety and stability of ammonia pump operation. Specifically, through multi-parameter monitoring and fault diagnosis models, recoverable and unrecoverable faults can be distinguished, reducing false positives and false negatives. Automatic interlocking control combined with frequency converter adjustment of outlet pressure allows the standby pump to seamlessly take over in the event of a main pump failure, avoiding drastic fluctuations in process pressure. The DCS system displays pump status, fault codes, trend graphs, and suggested measures, enabling operators to promptly assess fault evolution and implement maintenance measures. This reduces reliance on manual operation, lowers the risk of equipment damage, and ensures the continuous and stable operation of the ammonia pump system.

[0105] In some embodiments, to further address the second technical problem described in the background section, namely, "existing ammonia pump control systems lack systematic management and status monitoring, easily leading to delayed maintenance, insufficient fault warnings, and uneven equipment operation," in some embodiments of the present invention, an adaptive interlocking control method for ammonia pumps further includes:

[0106] Step 1: Record the fault trigger time, number of switching operations, and changes in operating conditions before and after switching, and store them in a relational database after being timestamped.

[0107] In some embodiments, when a fault or pump switching operation is detected, the backend server records the fault trigger event using the current time as a timestamp, and simultaneously collects operating parameters (flow rate, pressure, temperature, vibration, etc.) before and after the switching, recording the number of switching operations. The collected data is categorized and stored in a relational database in chronological order. Each record contains information such as fault time, number of switching operations, parameters before and after the switching, and fault type, facilitating subsequent querying and analysis. The fault trigger time is the specific time when the main ammonia pump or standby pump fails, used to record the sequence of fault occurrences. The number of switching operations is the number of switching operations between the main pump and the standby pump, used to analyze system stability and pump reliability. Changes in operating conditions before and after the switching are the changes in pump operating parameters before and after the switching, including flow rate, pressure, temperature, vibration, etc., used to assess the impact of the switching on the system. The timestamp is used to mark the specific time in the data record, ensuring that each record can be traced back to an accurate moment. A relational database, such as MySQL or PostgreSQL, is used to store structured data and manage pump operation and fault history data.

[0108] Step 2: Perform correlation analysis on the relational database based on association rules and output a fault troubleshooting list;

[0109] Step 3: Generate corresponding work orders based on the fault diagnosis list, automatically assign tasks to designated maintenance teams, and push spare parts requisition lists to the warehousing system;

[0110] In some embodiments, the backend server extracts fault trigger records, pump switching records, and historical operating parameter data from a relational database. It then analyzes the correlation between different parameters using association rule algorithms (such as Apriori and FP-Growth) to uncover fault patterns. For example, it might find that "historically, when the bearing temperature continuously exceeds 85 degrees Celsius and the vibration acceleration exceeds 5 mm / s², the probability of switching to the standby pump is 80%." Based on these rules, a fault troubleshooting list is generated, listing possible fault causes and their priority for inspection. Association rules are a data mining method used to discover correlations between different events or parameters in a database, such as "if the bearing temperature exceeds 90°C and the pump vibration exceeds the threshold, a bearing failure may occur." Correlation analysis analyzes the statistical relationships between different fault events, operating parameters, and operational behaviors to discover potential fault patterns or causal relationships. The fault troubleshooting list is an operation guide or troubleshooting step list generated based on the analysis results, used to guide maintenance personnel to quickly locate faults. Based on this, the fault troubleshooting list is organized into work orders, which include the fault items to be investigated, operation steps, priorities, and estimated completion times. Based on information such as work orders, maintenance team skill matrix, team's current workload, and geographical location, the system automatically assigns work orders to suitable teams. It also analyzes the spare parts required by the work order (such as pump bearings and seals), automatically generates a spare parts requisition list, and pushes the requisition request through the warehouse system, ensuring that maintenance teams can obtain the necessary materials in a timely manner when executing work orders. For example, work order generation: Work order number 20250822-01, content "Main pump bearing overheating troubleshooting," high priority, estimated completion time 2 hours. Automatic assignment: Team A has mechanical maintenance qualifications and is currently available; the work order is automatically assigned to Team A. Spare parts requisition: A list is generated, including 2 pump bearings and 1 seal, and a requisition request is submitted through the warehouse system. Automatic task assignment assigns work orders to suitable maintenance teams or personnel based on rules such as work orders, team skills, and workload, without manual intervention. The spare parts requisition list is an automatically generated material list based on the parts to be replaced or inspected according to the work order, used by the warehouse system to issue materials. A warehousing system is a system for managing inventory and material issuance, such as an ERP or MES system.

[0111] Step 4: Upload maintenance process data and original fault records in real time via mobile terminal to generate a maintenance effectiveness evaluation index;

[0112] Step 5: Automatically generate an electronic work order containing a timestamp, fault code, and handling operation, store it in the database, and add the fault characteristics to the machine learning training set.

[0113] In some embodiments, the mobile terminal is a smart device used by maintenance personnel in the maintenance team, such as a tablet, handheld terminal, or industrial mobile phone, to record and upload on-site maintenance data. Maintenance personnel use the mobile terminal on-site to perform operations according to work orders, simultaneously inputting operation steps and parameters in real time, and uploading maintenance process data and original fault records to the backend server via wireless network or industrial Ethernet. The server automatically calculates a maintenance effectiveness evaluation index based on the uploaded data. For example, it calculates the performance recovery ratio based on changes in pressure, flow, and vibration of the main and standby pumps before and after maintenance; it calculates the fault recurrence rate based on the time interval between faults triggered again after maintenance; and it generates an evaluation index from 0 to 100 based on a comprehensive analysis of various indicators to evaluate the effectiveness of maintenance measures. Maintenance process data refers to detailed information about maintenance operations, including operation steps, execution time, tools used, and operators. Original fault records refer to parameter data collected by the system when a fault is triggered, such as flow rate, pressure, temperature, and vibration—unprocessed raw data. The maintenance effectiveness evaluation index is a quantitative indicator used to evaluate the effectiveness of maintenance measures, typically calculated by combining parameters such as fault recovery time, changes in pump status before and after recovery, and recurrence rate. Based on this, upon completion of maintenance or during work order execution, the backend server automatically generates an electronic work order from the maintenance process data, fault codes, and operation steps, along with an event timestamp. The electronic work order is stored in a relational database or cloud database for subsequent querying, statistics, and auditing. Simultaneously, the real-time operating parameters, abnormal characteristics, repair results, and work order execution data corresponding to this fault are compiled into training samples and added to the machine learning training set to optimize the fault diagnosis model, enabling the model to learn new fault modes or verify the predictive accuracy of existing models. The electronic work order is a digitally recorded maintenance task order, including fault information, operation steps, timestamp, and personnel information, used by the system to track and manage maintenance activities. The timestamp records the specific time the event occurred, used to associate fault, operation, and operational data. The handling operation refers to the specific steps performed by maintenance personnel, such as switching to a standby pump, checking bearings, or adjusting valves. The database stores electronic work orders, operational data, and fault history. The machine learning training set is a dataset used to train or update the fault diagnosis model, including input features (flow rate, pressure, temperature, vibration, etc.) and corresponding labels (fault type, recoverable / unrecoverable fault, etc.). In summary, this application achieves centralized management of fault information through systematic data recording and storage, thereby improving the efficiency of maintenance decision-making; it quickly locates the root cause of problems and improves maintenance accuracy through correlation analysis and fault troubleshooting lists; it reduces manual intervention and improves maintenance response speed through automatic work orders and spare parts push; and it feeds maintenance and fault characteristic data back to the training set to support continuous optimization of machine learning models and achieve intelligent operation and maintenance.

[0114] The adaptive interlocking control method for an ammonia pump of the present invention further includes:

[0115] Based on the real-time operating parameters of the main ammonia pump and the standby pump, determine the health status level of the main ammonia pump and the standby pump;

[0116] The health status level is compared with the historical health status level to generate a comparison result; if the comparison result indicates that the ammonia pump has been downgraded, an early warning is generated and a planned maintenance is arranged.

[0117] In some embodiments, real-time health status assessment and historical comparison enable early identification, warning alerts, and planned maintenance of pump operating conditions, thereby improving pump reliability, extending service life, and optimizing maintenance efficiency. Specifically, daily, weekly, or monthly assessment cycles can be set. The backend server automatically reads pump sensor data (flow rate, pressure, vibration, temperature, etc.) and compares it with preset thresholds for each parameter to determine a health score for each parameter. Then, the health scores for each parameter are weighted to generate a total health score. Based on the total health score, a health level table is consulted to determine the health status level. Pre-stored historical health status levels are then queried and compared with the current health status level. If the health status level has deteriorated, a warning alert is generated, and planned maintenance is scheduled. The health status level is a comprehensive evaluation of the pump's operating status, typically categorized as "normal," "slightly abnormal," or "severely abnormal." The historical health status level is a record of the previous health status. Warning alerts are notifications issued by the system to remind operators of potential pump problems; planned maintenance is maintenance measures arranged based on the warnings.

[0118] The adaptive interlocking control method for an ammonia pump of the present invention further includes:

[0119] Record the cumulative operating time of the main ammonia pump and the standby pump separately. The cumulative operating time includes the main operating time and the standby time.

[0120] When the main operating time reaches the preset threshold and the main operating time is greater than or equal to the standby time, the main ammonia pump and the standby pump will be rotated.

[0121] During the rotation process, the same pressure smooth transition control method as that used for fault switching is adopted;

[0122] After the rotation is completed, the status of the main ammonia pump will be updated to standby pump.

[0123] In some embodiments, an automatic rotation strategy ensures balanced operation of the main pump and standby pump, extending the overall lifespan of the pump set; the pressure transition during rotation is smooth, avoiding the impact of transient pressure fluctuations on the system and ensuring process stability; and the operating status and time records are automatically managed, reducing the risk of human operation and improving the level of automation. Specifically, the cumulative operating time refers to the total time each pump has accumulated from startup to the current time, including the time the ammonia pump operates as the main ammonia pump (main operating time) and the time it operates as a standby pump (standby time). The preset threshold is a reference value for pump rotation after the operating time reaches a certain value, which can be set according to the pump's rated lifespan or maintenance strategy. Based on this, the background server continuously records the operating status and cumulative time of each pump. When the cumulative main operating time of the main ammonia pump reaches the preset threshold, and the main operating time is greater than or equal to the standby time, it is determined that a rotation operation is required: the standby pump is started, and the inverter output frequency and outlet valve opening are gradually adjusted according to the pressure smooth transition control method, so that the standby pump smoothly takes over the load of the main pump. After the rotation is completed, the status of the original main pump is updated to standby pump, and the status of the standby pump is updated to main pump. At the same time, the cumulative running time records of each pump are updated to provide data support for the next round of rotation or operation and maintenance decisions.

[0124] These embodiments improve maintenance efficiency and response speed. Specifically, through monitoring and analysis of faults, health status, and uptime, closed-loop management of pump unit status is achieved: automatically detecting abnormal or degraded pump units, generating work orders or early warning prompts, and executing rotation and pressure smoothing control to ensure reliable operation, timely maintenance, efficient allocation, and balanced lifespan of pump units, while supporting data accumulation for model optimization and subsequent decision-making.

[0125] In some embodiments, to further address the third technical problem described in the background section, namely, "existing ammonia pump controls are susceptible to unauthorized operations or abnormal behavior, potentially leading to risks in equipment operation and data management, resulting in low security and insufficient controllability," in some embodiments of the present invention, an adaptive interlocking control method for ammonia pumps further includes:

[0126] Step 1: Encrypt the real-time operating parameters and perform whitelist communication filtering through the industrial firewall deployed between the DCS system and the DCS system.

[0127] Step 2: Receive the operation records from the DCS system and add a timestamp to each record, including the operation time, operator ID, and equipment ID;

[0128] Step 3: Based on the pre-built operation rule base, perform pattern analysis on the operation behavior records to identify abnormal operation sequences, and generate a security alarm when an abnormal operation is detected.

[0129] Step four involves using a data isolation gateway deployed between the DCS system and the external network to perform protocol parsing and content review of the bidirectional data streams, and to verify the identity of operators based on a multi-level user access control system and digital certificates.

[0130] In some embodiments, the backend server encrypts the received real-time operating parameters, typically using symmetric encryption algorithms (such as AES) or asymmetric encryption algorithms (such as RSA) to prevent data theft or tampering during transmission. For example, the backend server encrypts the data using an AES key before sending it to the DCS system. The encrypted data is transmitted through an industrial firewall, which filters communication based on a preset whitelist, ensuring that only legitimate sensor data can enter the DCS system, thus achieving secure data transmission and access control. The industrial firewall is a network security device deployed between the backend server and the DCS system to control network traffic, allowing only devices or protocols verified through the whitelist to communicate. The whitelist refers to a list of device IDs or IP addresses allowed to communicate, preventing unauthorized devices or malicious traffic from entering the backend server. The whitelist-based communication filtering industrial firewall checks the data source and type of each transmission, allowing only data that meets the whitelist criteria to pass. Based on this, the backend server establishes a communication connection with the DCS system to receive operation records from the DCS system and adds a timestamp containing the operation time, operator ID, and device ID to each record. The operation behavior log refers to all operation information generated within the DCS system, including operation commands (such as starting / stopping pumps, adjusting valves), parameter modifications (such as pump speed, pressure thresholds), and alarm confirmations. Timestamps are added to each record with time information (accurate to milliseconds or seconds) to determine the chronological order of operations. Operator identification records the user ID or login account that performed the operation, clearly identifying the responsible party. Equipment identification records the specific equipment affected by the operation, such as the main ammonia pump or standby pump, facilitating subsequent analysis and traceability. The operation rule base is a set of predefined rules or patterns used to describe legitimate operation behavior sequences. Examples include: "The valve must be opened before starting the pump," and "Parameter adjustments must not exceed the set range." Based on this, the pre-built operation rule base is used to perform pattern comparison and sequence analysis on the received operation behavior logs. If certain operation logs violate the rules or exhibit an abnormal sequence (such as adjusting a pressure valve while the pump is off), the server determines it as an abnormal operation sequence. Abnormal operations that violate the rules are identified, and an alarm mechanism is triggered upon detection, sending alarm information to the monitoring interface, administrator, or security system. Pattern analysis refers to the backend server comparing and analyzing received operation records according to the rules of the operation rule base to determine whether the operation conforms to a preset pattern. The backend server works in conjunction with the deployed data isolation gateway to perform protocol parsing and content inspection on all data streams between the DCS system and the external network. At the same time, it uses multi-level access control and digital certificates to verify the identity of operators, ensuring the legitimacy of operations and preventing unauthorized access or abnormal operations.A data isolation gateway is a network security device deployed between the DCS system and the external network. It isolates the internal control system from the external network, preventing unauthorized access or malicious data from entering the internal system. Bidirectional data flow protocol parsing and content review refers to the gateway parsing data packets entering and leaving the DCS system, checking whether the data communication protocol conforms to preset specifications, and analyzing the data content to ensure there are no malicious commands or abnormal requests. A multi-level user access control system refers to the backend server's hierarchical management of operators, such as administrators, operators, and maintenance personnel. Each level of user can only access and operate within a permitted range of functions or data. Digital certificate verification of operator identity uses only digital certificates (such as X.509 certificates) to verify operator identity, ensuring that only legitimate users can perform operations.

[0131] These embodiments enhance the security and controllability of data transmission. Specifically, by employing multiple layers of security measures, including data encryption, firewall filtering, operational behavior auditing, pattern analysis, isolation gateways, and access authentication, an end-to-end industrial control system security protection and abnormal operation detection system is formed, thereby ensuring data integrity, operational traceability, and access legitimacy.

[0132] The above description is merely a selection of preferred embodiments of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to specific combinations of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. An adaptive interlocking control method for an ammonia pump, characterized in that, include: Multiple ammonia pumps are monitored in real time using radar level gauges, accelerometers, pressure transmitters, and temperature sensors. The multiple ammonia pumps include a main ammonia pump and a standby pump, in order to obtain the real-time operating parameters of the multiple ammonia pumps. The fault diagnosis model was used to screen faults based on the real-time operating parameters of the main ammonia pump, and the fault screening results were obtained. Based on the fault screening results, the main ammonia pump is interlocked and controlled. If the main ammonia pump is confirmed to be faulty, the standby pump is started within a preset time. The frequency of the inverter output at the ammonia pump outlet is adjusted to ensure a smooth transition of the outlet pressure. This includes: if the fault diagnosis model confirms that the main ammonia pump has an unrecoverable fault, continuously tracking the fault signal waveform within a preset time window, analyzing the synchronicity and trend correlation of multi-channel parameter changes, verifying the differences in signal attenuation characteristics and environmental interference modes, cross-comparing redundant sensor data with historical interference case databases, filtering transient abnormal pulses or random noise interference, and triggering the interlock protection mechanism after confirming the persistence of the fault; if no misjudgment is found, a start command is sent to the control system of the standby pump, and the standby pump preloads the operating parameters of the main ammonia pump before shutdown; the power supply to the drive motor of the main ammonia pump is cut off; the drive motor of the standby pump is soft-started according to a preset acceleration curve, and the outlet valve is slowly opened to the fully open position; the status of the currently running ammonia pump is updated to standby pump. The start / stop status of the ammonia pump is displayed on the DCS system's operating interface. When a fault is confirmed in the main ammonia pump, the fault code, suggested measures, and real-time trend graph are displayed.

2. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, The method employs radar level gauges, accelerometers, pressure transmitters, and temperature sensors to perform real-time multi-parameter monitoring of multiple ammonia pumps. These multiple ammonia pumps include main ammonia pumps and standby pumps. The method obtains real-time operating parameters of these multiple ammonia pumps, including: The inlet flow rates of the main ammonia pump and the standby pump are monitored in real time using radar level gauges. Vibration data of the main ammonia pump and the standby pump are collected by an accelerometer. By analyzing the vibration data, mechanical faults of the main ammonia pump and the standby pump are identified. Temperature data is collected by monitoring the bearing temperature and motor stator temperature of the main ammonia pump and the standby pump using temperature sensors. The pressure transmitter is used to monitor the inlet and outlet pressures of the main ammonia pump and the standby pump in real time, and abnormal pressure fluctuations exceeding the preset pressure threshold are used as the basis for fault diagnosis.

3. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, The fault diagnosis model is used to screen faults based on the real-time operating parameters of the main ammonia pump, including: The historical operating parameters of the main ammonia pump are obtained, and signal features are extracted using time-domain analysis and frequency-domain analysis. The historical operating parameters are then subjected to feature dimensionality reduction and divided into transient parameters and steady-state parameters. Historical fault data of the main ammonia pump is obtained, and transfer learning optimization is performed based on the historical fault data. The weight allocation of key features is strengthened by combining the attention mechanism, and a machine learning model is constructed. For steady-state parameters, a sliding window statistical threshold is used, and a delayed confirmation mechanism is introduced for transient parameters. By fusing multi-parameter anomaly confidence through Bayesian inference, the adjusted machine learning model is confirmed as a fault diagnosis model. Based on the real-time operating parameters of the main ammonia pump, the fault diagnosis model is used to perform fault diagnosis. Identify abnormal parameters in the real-time operating parameters of the main ammonia pump, analyze the causal relationship between abnormal parameters, pinpoint the root cause of the fault, and generate graded alarm signals based on the scope and urgency of the fault.

4. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, The DCS system's operating interface displays the start / stop status of the ammonia pump. When a fault is confirmed in the main ammonia pump, it displays a fault code, suggested measures, and a real-time trend graph, including: A dynamic indicator module for ammonia pumps is embedded in the operating interface of the DCS system to display the real-time operating parameters of the main ammonia pump and the standby pump. Based on the fault codes, graded alarm signals, and preset knowledge base, determine the recommended measures and display them on the DCS system's operation interface; A layered window pops up on the DCS system's operating interface. The left side displays the fault code and text description, the right side embeds a real-time trend chart, and overlays historical data curves under the same working conditions to help judge the fault evolution trend. The bottom displays handling steps buttons with priority sorting. When a recoverable fault is detected in the main ammonia pump, a preset first-level alarm function is triggered, which activates a yellow rotating light and a low-frequency buzzer. When an unrecoverable fault is detected in the main ammonia pump, a second-level alarm function is activated, which activates a red flashing light and an alarm sound.

5. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, Also includes: Record the fault trigger time, number of switching times, and changes in operating conditions before and after switching, and store them in a relational database after being timestamped; Based on association rules, correlation analysis is performed on the relational database to output a fault troubleshooting list; Based on the fault diagnosis list, a corresponding work order is generated, the task is automatically assigned to the designated maintenance team, and the spare parts requisition list is pushed to the warehousing system. The maintenance process data and original fault records are uploaded in real time via mobile terminals to generate a maintenance effectiveness evaluation index. Automatically generate electronic work orders containing timestamps, fault codes, and handling operations, store them in the database, and add the fault characteristics to the machine learning training set.

6. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, Also includes: Based on the real-time operating parameters of the main ammonia pump and the standby pump, determine the health status level of the main ammonia pump and the standby pump; The health status level is compared with the historical health status level to generate a comparison result; If the comparison results indicate that the ammonia pump has degraded, an early warning will be generated and a planned overhaul will be scheduled.

7. The adaptive interlocking control method for an ammonia pump according to claim 1, characterized in that, Also includes: Record the cumulative operating time of the main ammonia pump and the standby pump separately. The cumulative operating time includes the main operating time and the standby time. When the main operating time reaches the preset threshold and the main operating time is greater than or equal to the standby time, the main ammonia pump and the standby pump will be rotated. During the rotation process, the same pressure smooth transition control method as that used for fault switching is adopted; After the rotation is completed, the status of the main ammonia pump will be updated to standby pump.

Images