Marine engine room monitoring and alarming method and system based on edge calculation

Abstract

The invention relates to the field of ship safety, and discloses a ship engine room monitoring and alarming method and system based on edge computing, and the method comprises the following steps: firstly, collecting environment parameters and equipment operation parameters in a ship engine room in real time; secondly, performing standardization processing on the acquired environment parameters and the operation parameters, and taking the processed parameters as monitoring data; then, sending the monitoring data to an edge calculation center for preliminary analysis so as to screen out preliminary abnormal parameters corresponding to data exceeding a threshold range; finally, a screening list is determined based on the preliminary abnormal parameters, sorting is carried out, and the priority of the screening list is comprehensively determined by combining the number of times of entering the screening list in unit time, the difference value between the real-time parameter value and the median of the threshold value range and whether the associated preliminary abnormal parameters exist or not during sorting; through the scheme, the potential risk of the cabin is analyzed, the response efficiency is improved, and the probability of fault hazard expansion is reduced.

Description

Technical Field

[0001] The present invention relates to the field of ship safety, and in particular to a ship engine room monitoring and alarm method and system based on edge computing. Background Art

[0002] The ship's engine room is a key monitoring area for ship safety, and most ship failures originate from the engine room. Although existing ship engine rooms are equipped with a variety of sensors to monitor safety, there are still some problems in actual use.

[0003] For example, during normal operation, sudden acceleration or other factors can cause the temperature of engine room equipment to suddenly increase. Sensors with a single threshold warning will then sound an alarm, resulting in a false alarm, which could potentially affect the ship's daily operation. Furthermore, the engine room presents potential failure risks during daily adaptation. Existing monitoring systems, relying on a single threshold comparison, struggle to detect potential failure risks in indicators that do not exhibit obvious anomalies. Furthermore, the cause of an obviously abnormal indicator may be due to slight changes in other related indicators. Clearly, existing monitoring systems struggle to detect and respond promptly, especially if the maintenance priority of monitored indicators is misjudged. This can easily lead to a delayed response to ship failures and exacerbated damage. Summary of the Invention

[0004] The purpose of the present invention is to provide a ship engine room monitoring and alarm method and system based on edge computing to solve the above technical problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A ship engine room monitoring and alarm method based on edge computing, the method specifically includes the following steps:

[0007] S1. Real-time collection of environmental parameters and equipment operating parameters in the ship's engine room;

[0008] S2. Standardize the collected environmental parameters and operating parameters and use them as monitoring data;

[0009] S3. Send the monitoring data to the edge computing center for preliminary analysis to filter out preliminary abnormal parameters corresponding to data that exceeds the threshold range;

[0010] S4. Determine and sort the screening list based on preliminary abnormality parameters. The priority of the screening list is determined based on the number of entries per unit time, the difference between the real-time parameter value and the median of the threshold range, and whether there are any related preliminary abnormality parameters. Through real-time parameter collection, standardized processing, preliminary edge computing analysis, and multi-dimensional sorting, accurate screening and prioritization of engine room abnormalities are achieved. Compared with traditional single-threshold monitoring, it can effectively distinguish between substantial abnormalities and short-term fluctuations, avoiding false alarms that interfere with ship operations. At the same time, critical abnormalities are prioritized, improving fault response efficiency and reducing the risk of fault expansion.

[0011] As a further technical solution, methods for determining the priority of the screening list include:

[0012] S41, receive the preliminary abnormal parameter list from the edge computing center, A={P1, P2, ..., P n};

[0013] S42, for each parameter P i Parallel execution:

[0014] Call the frequency statistics module to calculate the frequency factor F i ;

[0015] Call the deviation calculation module to calculate the deviation factor D i ;

[0016] Query the pre-set association rule library to determine the correlation factor C i ;

[0017] S43, F i 、D i and C i Input into the scoring model and output the score PS of each parameter i ;

[0018] S44, press A to PS i Output the priority queue after sorting in descending order.

[0019] As a further technical solution, the frequency statistics module is called to calculate the frequency factor F i The specific methods include:

[0020] Count the cumulative number of times the current preliminary abnormal parameter is marked as preliminary abnormal within the preset time period T and perform normalization processing;

[0021] The specific formula is: i =min(N t / N max ,1); where N t is the number of abnormalities within the preset time period T, N maxThe preset maximum count limit.

[0022] As a further technical solution, the deviation calculation module is called to calculate the deviation factor D i The methods include:

[0023] Calculate the relative deviation intensity of the real-time value based on the median value of the preset threshold range;

[0024] The specific formula is: in, P imax 、P imin are the upper and lower thresholds of the current preliminary abnormal parameters respectively; P iX is the real-time value of the current preliminary abnormal parameter, P iM is the median value of the current preliminary abnormal parameters.

[0025] As a further technical solution, the correlation factor C i The method to obtain is:

[0026]

[0027] Among them, M is the number of association rules currently triggered, w i is the weight of the i-th rule, M max is the maximum number of association rules threshold.

[0028] As a further technical solution, the expression of the scoring model in S43 is:

[0029]

[0030] Among them, Base is the basic anomaly score, EN is the dynamic enhancement coefficient, and Im is the correlation impact score. is the adjustment factor.

[0031] As a further technical solution, the calculation formula of the current preliminary abnormal parameter basic abnormal score is: Base = ω p *[θ*D i +(1-θ)*F i ]*K(P i );

[0032] Among them, K(P i ) is the abnormality level function, θ is the dynamic weight, ω p Preset weights for parameters;

[0033] Among them, η is the enhancement coefficient, which defaults to 1.2 and can be adjusted according to the importance of the parameter; t is the abnormal duration of the initial abnormal parameter, t max is a preset maximum duration threshold;

[0034] Im=ω c *C i *K(P i ); where ω c is the association weight (set by experts and normalized to 0-0.5, such as 0.3).

[0035] As a further technical solution, the K(P i ) is:

[0036] When the parameter real-time value P iX Exceeding the preset threshold range [P imin , P imax ], then

[0037] A ship engine room monitoring and alarm system based on edge computing, which is used to execute the ship engine room monitoring and alarm method based on edge computing.

[0038] Beneficial effects of the present invention:

[0039] (1) The present invention constructs a multi-dimensional scoring model by introducing frequency factors, deviation factors and correlation factors, which effectively solves the problem of false alarms caused by single threshold warnings in the background technology. In traditional methods, abnormalities triggered by short-term fluctuations such as sudden acceleration are prone to false alarms. However, the present invention counts the abnormal frequency within a preset time period and normalizes it, combines the relative deviation analysis between the real-time value and the median threshold value, and then associates it with the abnormal conditions of other parameters for comprehensive judgment. It can accurately distinguish between substantial abnormalities and short-term fluctuations, greatly reduce the false alarm rate, and avoid unnecessary interference with the daily operation of the ship.

[0040] (2) The present invention addresses the problem that potential fault risks are difficult to detect in the background technology, and realizes effective identification of potential risks through multi-factor collaborative evaluation. Specifically, since the existing monitoring system relies on a single threshold comparison, it is powerless to detect indicators that are not obviously abnormal but have hidden dangers. The present invention calculates the degree of deviation of the real-time value relative to the median threshold value through the deviation factor, and combines the association rule library to mine the potential correlation between parameters. It can capture subtle abnormal changes in indicators and their associated effects, discover potential fault risks in advance, and buy time for fault prevention.

[0041] (3) The present invention solves the problem of misjudgment of maintenance priority of abnormal indicators in the background technology, and ensures timely handling of faults through accurate sorting. Compared with the traditional system, it is difficult to judge the urgency of abnormal indicators, which may lead to delayed response to key faults. The present invention inputs frequency, deviation, and correlation factors into the scoring model to obtain parameter scores, and arranges them in descending order of scores to form a priority queue, so that key abnormalities that affect ship safety are given priority, avoiding the aggravation of fault damage, and improving the safety monitoring and emergency response efficiency of the ship's engine room. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will be further described below with reference to the accompanying drawings.

[0043] Figure 1 A diagram showing the steps of the method of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] See also Figure 1 As shown, the present invention is a ship engine room monitoring and alarm method based on edge computing, which specifically includes the following steps:

[0046] S1. Real-time collection of environmental parameters and equipment operating parameters in the ship's engine room. These parameters include engine room temperature, pressure, and equipment speed, which are collected in real time through sensors. These parameters are acquired using hardware devices such as temperature sensors and pressure sensors deployed in the engine room.

[0047] S2. Standardize the collected environmental and operating parameters and use them as monitoring data. Standardization includes preprocessing the collected raw parameters, including removing data noise (e.g., filtering out sensor transient errors), unifying dimensions, and normalizing (e.g., mapping parameter values ​​to a range of 0-1) to ensure consistency in subsequent analysis.

[0048] S3. The monitoring data is sent to the edge computing center for preliminary analysis to screen out preliminary abnormal parameters corresponding to data that exceeds the threshold range. The edge computing center compares the monitoring data with the preset threshold range and marks any data that exceeds the threshold range as a preliminary abnormality. The edge computing center is deployed locally on the ship and can quickly process data, such as reducing cloud transmission delays. By comparing with the preset threshold, it can preliminarily screen out parameters that exceed the range, laying the foundation for subsequent detailed analysis.

[0049] S4. Determine the screening list based on the preliminary abnormal parameters and sort it. When sorting, the priority of the screening list is determined comprehensively based on the number of times the screening list is entered per unit time, the difference between the real-time parameter value and the median of the threshold range, and whether there are related preliminary abnormal parameters.

[0050] In this embodiment, through real-time parameter collection, standardized processing, preliminary edge computing analysis, and multi-dimensional sorting, accurate screening and prioritization of engine room anomalies are achieved. Compared with traditional single-threshold monitoring, this can effectively distinguish between substantial anomalies and short-term fluctuations, avoiding false alarms that interfere with ship operations. At the same time, critical anomalies are handled first, improving fault response efficiency and reducing the risk of fault escalation.

[0051] By introducing frequency factors, deviation factors and correlation factors to construct a multi-dimensional scoring model, the false alarm problem caused by a single threshold warning in the background technology is effectively solved. In the traditional method, anomalies triggered by short-term fluctuations such as sudden acceleration are prone to false alarms. The present invention counts the abnormal frequency within a preset time period and normalizes it, combines the relative deviation analysis between the real-time value and the median threshold, and then associates it with the abnormal conditions of other parameters for comprehensive judgment. It can accurately distinguish between substantial abnormalities and short-term fluctuations, greatly reduce the false alarm rate, and avoid unnecessary interference with the daily operation of the ship.

[0052] Methods for prioritizing screening lists include:

[0053] S41, receive the preliminary abnormal parameter list from the edge computing center, A={P1, P2, ..., P n}; The abnormal parameter set output by the edge computing center, obtained by: edge computing center transmission;

[0054] S42, for each parameter P i Parallel execution:

[0055] Call the frequency statistics module to calculate the frequency factor F i ;

[0056] Call the deviation calculation module to calculate the deviation factor D i ;

[0057] Query the pre-set association rule library to determine the correlation factor C i ; respectively reflect the abnormal frequency, deviation degree, and correlation impact; parallel execution means calculating each parameter in List A simultaneously, rather than one by one, which can improve processing efficiency. For example, factor calculation of temperature and pressure parameters is performed simultaneously, which is suitable for scenarios where multiple parameters are abnormal at the same time;

[0058] S43, F i 、D i and C iInput into the scoring model and output the score PS of each parameter i ;

[0059] S44, press A to PS i Output the priority queue after sorting in descending order.

[0060] In this embodiment, by parallel calculation of frequency, deviation, and correlation factors and building a scoring model, accurate ranking of abnormal parameter priorities is achieved, thus avoiding the one-sidedness of traditional methods that sort by abnormality occurrence order or a single indicator. This ensures that key abnormalities that affect ship safety, such as core equipment failures, receive priority attention, improving the targeted nature of fault handling.

[0061] At the same time, in response to the problem that potential fault risks are difficult to detect in background technologies, effective identification of potential risks is achieved through multi-factor collaborative evaluation; specifically, since the existing monitoring system relies on a single threshold comparison, it is powerless for indicators that are not obviously abnormal but have hidden dangers. The present invention calculates the degree of deviation of the real-time value relative to the median of the threshold through the deviation factor, and combines the association rule library to mine the potential correlation between parameters. It can capture subtle abnormal changes in indicators and their associated impacts, discover potential fault risks in advance, and buy time for fault prevention.

[0062] Call the frequency statistics module to calculate the frequency factor F i The specific methods include:

[0063] Count the cumulative number of times the current preliminary abnormal parameter is marked as preliminary abnormal within the preset time period T and perform normalization processing;

[0064] The purpose of normalization is to convert N t / N max The result is limited to 0-1 to avoid N t Much larger than N max Lead to F i Too large, such as N t =10, N max =5, F i = 1 instead of 2, ensuring that the same i and C i The preset time period T is set according to the characteristics of the equipment, such as 5 minutes for the easily fluctuating parameter T and 15 minutes for the stable parameter T, to balance real-time performance and accuracy;

[0065] The specific formula is: i =min(N t / N max ,1); where N t is the number of abnormalities within the preset time period T, N maxFor example, if a temperature parameter triggers an exception three times within 10 minutes, N max =5→F i =3 / 5=0.6. F i The range is 0-1, with larger values ​​indicating more frequent anomalies.

[0066] In this embodiment, by counting and normalizing the abnormality frequency within a preset time period, occasional abnormalities are effectively distinguished from continuous abnormalities, thereby avoiding false alarms caused by transient sensor fluctuations, such as a short-term temperature increase caused by sudden acceleration of the device, so that frequent real abnormalities receive more attention; the calculation formula F i =min(N t / N max ,1) The number of abnormalities N within the preset time period T t Specifically, it is: the cumulative number of times the parameter is marked as a preliminary abnormality within the preset time period T. The acquisition method is: the edge computing center counts the number of abnormal markings within T, such as the number of times the temperature exceeds the threshold within 10 minutes; N max The maximum count limit is set by experts based on the characteristics of the equipment, such as the core equipment temperature N max Set it to 5 times to prevent interference from high-frequency sensor failure.

[0067] Call the deviation calculation module to calculate the deviation factor D i The methods include:

[0068] Calculate the relative deviation intensity of the real-time value based on the median value of the preset threshold range;

[0069] The specific formula is: in, P imax 、P imin are the upper and lower thresholds of the current preliminary abnormal parameters respectively; P iX is the real-time value of the current preliminary abnormal parameter, P iM is the median value of the current preliminary abnormal parameters, P imax -P imin is the threshold interval width (as the normalized denominator); Example: If the oil pressure threshold range is [80,120]kPa→P iM =100kPa, real-time value

[0070] In this embodiment, the degree of deviation of the real-time value is quantified based on the median value of the threshold, breaking through the traditional limitation of only judging whether the threshold is exceeded, thereby capturing the degree to which the parameter deviates from the normal range, such as slight exceeding of the threshold and severe exceeding of the threshold, providing a basis for identifying potential risks. For example, although the parameter exceeds the threshold but the deviation is small, it may be a minor abnormality; if the deviation is large, it may be a serious fault; the relative deviation intensity refers to the deviation ratio of the real-time value relative to the median value of the threshold rather than the absolute difference, which can eliminate the influence of differences in the threshold ranges of different parameters.

[0071] Correlation factor C i The method to obtain is:

[0072]

[0073] Among them, M is the number of association rules currently triggered, w i is the weight of the i-th rule, M max is the maximum number of association rules threshold.

[0074] In this embodiment, the potential connections between parameters are mined through the association rule library, and the single parameter anomaly is analyzed in the overall system to avoid missed judgments caused by isolated judgments. For example, a parameter anomaly may be the result of an associated parameter failure, which improves the ability to identify systemic failures. For example, oil pressure anomaly may constitute a systemic failure together with associated temperature and flow anomalies. The calculation formula is: In the example, M is the number of currently triggered association rules, which can be obtained by querying the preset association rule library. For example, if the oil pressure anomaly is associated with the temperature anomaly, there is one rule. If both are triggered at the same time, M = 1. i According to the historical fault data setting, such as oil pressure-temperature correlation weight is 0.3, oil pressure-flow correlation weight is 0.2; M max is the maximum association rule number threshold, which is set according to the complexity of the ship system, such as 3;

[0075] The association rule base is a database that stores the association relationship between parameters. For example, if the cooling water flow is abnormal, the water pump speed is associated with abnormality. If the CO concentration is abnormal, the ventilation equipment is associated with abnormality. The rules are preset by experts based on the operation logic of the ship's engine room equipment and historical fault cases. The min function is used to limit C i The maximum value is 1 to avoid overflow of the correlation factor due to M being too large. Example:

[0076] The association rule base is constructed by the following steps:

[0077] (a) Collect historical fault data and extract parameter anomaly correlation patterns; for example, when oil pressure is abnormal, 90% of the time it is accompanied by cooling water temperature abnormality;

[0078] (b) Calculate the rule weight w based on the association probability i , such as w i= P (related parameter abnormality | current parameter abnormality) = the number of times the main parameter and the related parameter are abnormal at the same time / the total number of times the main parameter is abnormal;

[0079] (c) The rule format is defined as a triple: <main parameter, associated parameter, weight>, example: <oil pressure, cooling water temperature, 0.85>;

[0080] (d) Rule base update mechanism: The weights are optimized every quarter based on newly added fault data, and new rules must be verified through simulation.

[0081] The expression of the scoring model in S43 is:

[0082]

[0083] Among them, Base is the basic anomaly score, EN is the dynamic enhancement coefficient, and Im is the correlation impact score. To adjust the factor, calibrate it according to the operating results. For example, it is set to 1 in the initial stage and adjusted to 0.9 according to the false alarm rate in the later stage.

[0084] In this embodiment, frequency, deviation, correlation factor, and dynamic enhancement coefficient are integrated to output a quantitative score, making the anomaly priority sorting more objective, thereby avoiding the arbitrariness of traditional subjective priority judgments and ensuring that the scoring results are consistent with the actual severity and urgency of the anomaly. For example, anomalies with high frequency, high deviation, and strong correlation have higher scores and are processed first.

[0085] Calculation formula In the equation, Base reflects the basic severity of the abnormality, EN increases with the duration of the abnormality, and if the abnormality persists, it needs to be treated first. Im reflects the influence of the associated parameters. The three are combined to form Calibration, final output PS i ;PS i The higher the value, the higher the urgency and severity of the parameter anomaly, and the higher it is ranked in the priority queue.

[0086] The calculation formula for the current preliminary abnormal parameter basic abnormal score is: Base = ω p *[θ*D i +(1-θ)*F i ]*K(P i );

[0087] Among them, K(P i ) is the abnormality level function, which is nonlinearly scored according to the degree to which the real-time value of the parameter exceeds the threshold, and is divided into warning area and danger area. θ is the dynamic weight, which defaults to 0.7. When the frequency factor F iWhen it is greater than 0.8 (i.e., frequent abnormalities in a short period of time), θ is adjusted to 0.4 to give more emphasis to the frequency. The reason is that the adjustment logic of the dynamic weight θ is based on the statistical data of ship engine room failures: when F i When θ>0.8, frequent anomalies are more likely to be caused by equipment degradation, i.e., non-instantaneous fluctuations, so the real-time deviation weight θ is reduced to 0.4;

[0088] ω p Preset weights for parameters (determined by expert experience and normalized to between 0 and 1), for example: core equipment temperature: 0.9, lubricating oil pressure: 0.85, cooling water flow: 0.75, cabin ambient CO concentration: 0.6.

[0089] Among them, K(P i ) is the abnormality level function, θ is the dynamic weight, ω p Preset weights for parameters;

[0090] Where η is the enhancement coefficient, which defaults to 1.2 and can be adjusted according to the importance of the parameter and determined by historical data analysis. t is the abnormal duration of the initial abnormal parameter, which starts from the time the parameter is first marked as abnormal. max is a preset maximum duration threshold;

[0091] Im=ω c *C i *K(P i ); where ω c is the association weight (set by experts and normalized to 0-0.5, such as 0.3).

[0092] In this embodiment, the constituent factors of the scoring model are refined to quantify the impact of each factor and improve the interpretability of the model. i ) to make the scoring more relevant to the actual scenario, such as giving more weight to the frequency of high-frequency anomalies and amplifying the impact of severe anomalies;

[0093] The K(P i ) is:

[0094] When the parameter real-time value P iX Exceeding the preset threshold range [P imin , P imax ], then

[0095] In this embodiment, the severity of the parameter exceeding the threshold is quantified by an exponential function, reflecting the nonlinear relationship between the abnormality degree and the fault risk. For example, the more the parameter exceeds the threshold, the faster the risk increases, thereby avoiding treating minor abnormalities and serious abnormalities equally, giving serious abnormalities a higher score and ensuring that they are handled first. In the expression The reason for using the exponential function is that the abnormal risk of ship engine room equipment is nonlinearly related to the degree of parameter deviation. For example, the risk of exceeding the upper limit by 10°C is much higher than that of exceeding the upper limit by 5°C by two times. The exponential function can amplify the impact of serious abnormalities. For example, the more the upper limit is exceeded, the greater the risk of K(P i ) grows faster.

[0096] A ship engine room monitoring and alarm system based on edge computing, which is used to execute the ship engine room monitoring and alarm method based on edge computing.

[0097] It should be noted that the calculation formulas and various parameters involved in the calculations in the present invention have been dimensionally processed in advance, and the process of dimensionless processing is well known in the industry and will not be described here.

[0098] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A ship engine room monitoring and alarm method based on edge computing, characterized in that: The method specifically comprises the following steps: S1. Real-time collection of environmental parameters and equipment operating parameters in the ship's engine room; S2. Standardize the collected environmental parameters and operating parameters and use them as monitoring data; S3. Send the monitoring data to the edge computing center for preliminary analysis to filter out preliminary abnormal parameters corresponding to data that exceeds the threshold range; S4. Determine the screening list based on the preliminary abnormal parameters and sort it. When sorting, the priority of the screening list is determined comprehensively based on the number of times the screening list is entered per unit time, the difference between the real-time parameter value and the median of the threshold range, and whether there are related preliminary abnormal parameters.

2. The ship engine room monitoring and alarm method based on edge computing according to claim 1 is characterized in that: Methods for prioritizing screening lists include: S41, receive the preliminary abnormal parameter list from the edge computing center, A={P1, P2, ..., P n }; S42, for each parameter P i Parallel execution: Call the frequency statistics module to calculate the frequency factor F i ; Call the deviation calculation module to calculate the deviation factor D i ; Query the pre-set association rule library to determine the correlation factor C i ; S43, F i 、D i and C i Input into the scoring model and output the score PS of each parameter i ; S44, press A to PS i Output the priority queue after sorting in descending order.

3. The ship engine room monitoring and alarm method based on edge computing according to claim 2 is characterized in that: Call the frequency statistics module to calculate the frequency factor F i The specific methods include: Count the cumulative number of times the current preliminary abnormal parameter is marked as preliminary abnormal within the preset time period T and perform normalization processing; The specific formula is: i =min(N t / N max ,1); where N t is the number of abnormalities within the preset time period T, N max The preset maximum count limit.

4. The ship engine room monitoring and alarm method based on edge computing according to claim 2 is characterized in that: Call the deviation calculation module to calculate the deviation factor D i The methods include: Calculate the relative deviation intensity of the real-time value based on the median value of the preset threshold range; The specific formula is: in, P imax 、P imin are the upper and lower thresholds of the current preliminary abnormal parameters respectively; P iX is the real-time value of the current preliminary abnormal parameter, P iM is the median value of the current preliminary abnormal parameters.

5. The ship engine room monitoring and alarm method based on edge computing according to claim 2 is characterized in that: Correlation factor C i The method to obtain is: Among them, M is the number of association rules currently triggered, w i is the weight of the i-th rule, M max is the maximum number of association rules threshold.

6. The ship engine room monitoring and alarm method based on edge computing according to claim 2 is characterized in that: The expression of the scoring model in S43 is: Among them, Base is the basic anomaly score, EN is the dynamic enhancement coefficient, and Im is the correlation impact score. is the adjustment factor.

7. The ship engine room monitoring and alarm method based on edge computing according to claim 6 is characterized in that: The calculation formula for the current preliminary abnormal parameter basic abnormal score is: Base = ω p *[θ*D i +(1-θ)*F i ]*K(P i ); Among them, K(P i ) is the abnormality level function, θ is the dynamic weight, ω p Preset weights for parameters; Among them, η is the enhancement coefficient, which defaults to 1.2 and can be adjusted according to the importance of the parameter; t is the abnormal duration of the initial abnormal parameter, t max is a preset maximum duration threshold; Im=ω c *C i *K(P i ); where ω c is the association weight.

8. The ship engine room monitoring and alarm method based on edge computing according to claim 7 is characterized in that: The K(P i ) is: When the parameter real-time value P iX Exceeding the preset threshold range [P imin , P imax ], then 9. A ship engine room monitoring and alarm system based on edge computing, characterized in that: The alarm system is used to execute the ship engine room monitoring and alarm method based on edge computing as described in any one of claims 1-8.

Images