Ship refrigerated container cluster redundancy control method and system

Abstract

The invention provides a ship refrigerated container cluster redundancy control method and system, and the method comprises the steps: deploying a plurality of temperature sensors in each refrigerated container, and obtaining high-reliability environment temperature information through employing a multi-source data fusion technology; monitoring the running state of the local main controller of the refrigerated container in real time, judging whether the local main controller fails or not, and sending out an alarm signal; the redundant system monitors and stores all key temperature data of the refrigerated container in real time, and performs simple preprocessing to form a historical temperature database; comparing the control efficiency of the local main / backup controller and the control efficiency of the redundant system, and deciding an output signal weight; an accumulated historical database is utilized to dynamically adjust and optimize an adaptive fuzzy PID control rule in the redundant system; and collecting and counting models of local main / standby controllers and comparison and selection results of controller states and efficiency, and uploading the comparison and selection results to the cloud to construct an equipment efficiency knowledge base, so that faults can be accurately identified in a short time, the global temperature field characterization precision is improved, and unified regulation and control of containers of multiple brands are realized.

Description

Technical Field

[0001] This invention belongs to the technical field of refrigerated container fault monitoring and control, specifically relating to a redundancy control method and system for ship refrigerated container clusters. Background Technology

[0002] With the rapid development of global trade and cold chain logistics, refrigerated containers play a crucial role in maritime transport. The quality and safety of transported goods are highly dependent on the internal environment of the container, especially the precise and reliable control of temperature. Once the temperature deviates from the set strict range, the goods are highly susceptible to spoilage and deterioration within a short period of time, resulting in significant economic losses. Existing temperature control systems for refrigerated containers mainly rely on their Local Embedded Controllers (LECs). The controller operation modes are divided into local manual control and remote control via a central control console. The former requires operators to frequently visit the container to set or adjust parameters, which is inefficient and poses safety risks; while the latter can send setting commands remotely via a central control console through wired / wireless means, its control effect is essentially highly dependent on the performance and reliability of each individual LEC device.

[0003] However, existing technologies have many problems in ensuring the reliability of container cluster operations. The primary problem lies in the single point of failure risk of the primary container controller (LEC) and the serious inadequacy of existing redundancy mechanisms: the average annual failure rate of the LEC is about 1.7%, and the failure of a single unit means the loss of the temperature control function of the container it controls, directly leading to cargo damage. Although some containers are designed with backup controllers, their application has fundamental limitations: to balance costs, the performance of backup controllers is usually lower than that of the primary LEC, resulting in the inability to maintain the same level of temperature control after taking over, and the risk of cargo damage still exists during the takeover period; using backup solutions with chips of the same performance as the primary LEC is expensive, and the cost-effectiveness is extremely low due to long-term idleness; existing redundancy solutions essentially adopt local primary / backup switching logic for individual units, lacking an economical and highly available redundancy architecture at the cluster level.

[0004] On the other hand, existing cluster environmental monitoring and control capabilities also have significant shortcomings. Although current monitoring systems can collect and display container monitoring parameters, their measurement accuracy and reliability are often limited, failing to obtain high-precision environmental data that reflects the true situation. More critically, existing systems generally suffer from the problem of "strong monitoring but weak control" or even "monitoring without control": most systems only provide monitoring and alarm functions, and control commands still require manual on-site operation; the few systems that support remote control console commands still require issuing commands, heavily relying on the execution capabilities of individual LECs to achieve control actions. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for redundancy control of ship refrigerated container clusters, addressing the aforementioned problems.

[0006] The embodiments of this application are implemented as follows:

[0007] This application provides a method for redundancy control of refrigerated container clusters on ships, characterized by the following steps:

[0008] Step a: Deploy multiple temperature sensors inside each refrigerated container and use multi-source data fusion technology to obtain ambient temperature information;

[0009] Step b: Monitor the operating status of the local main controller of the refrigerated container in real time, determine whether it has failed, and issue an alarm signal;

[0010] Step c: The redundant system monitors and stores all key temperature data of refrigerated containers in real time, and performs simple preprocessing to form a historical temperature database.

[0011] Step d: Compare the control performance of the local primary or backup controller with that of the redundant system, and decide on the output signal weight;

[0012] Step e: Utilize the accumulated historical database to dynamically adjust and optimize the adaptive fuzzy PID control rules within the redundant system;

[0013] Step f: Collect and statistically analyze the local main / backup controller models, controller status, and performance comparison results, and upload them to the cloud to build a device performance knowledge base.

[0014] In some alternative implementations, step a includes the following specific details:

[0015] Deploy n temperature sensors in a refrigerated container and output the global temperature field status through a triple-weighted fusion mechanism:

[0016] Step a1: Calculate the initial arithmetic mean temperature using the following formula.

[0017] Among them, T i This indicates the measured temperature value of each temperature sensor.

[0018] Step a2, based on temperature deviation and rate of change Calculate dynamic weights

[0019]

[0020] in, Let represent the dynamic weights of the sensor in the k-th iteration. K represents the weighted average temperature calculated from sensor data at the k-th iteration. i K represents the position sensitivity coefficient. i ∈[0.6, 1.2], A i A represents the temperature change suppression coefficient. i ∈[0.3, 0.8],

[0021] Average value updated based on dynamic weights

[0022]

[0023] And through convergence conditions Continue until convergence, or until the number of iterations k≥3 controls the computational precision;

[0024] Step a3: When the temperature exceeds the preset threshold, the segmented adjustment of the risk weight W is activated. r,i :

[0025]

[0026] Where, Θ high Indicates the high temperature threshold, Θ low Indicates the low temperature threshold;

[0027] When T i When the temperature is above 5℃, the square of the weight increases, amplifying the risk of overheating; when T i At temperatures below 2°C, the weighted power function decreases, mitigating the risk of overcooling.

[0028] Step a4, Triple Fusion Output:

[0029] W final，i =0.5·W s，i +0.3·W d，i +0.2·W r，i

[0030] Among them, W s，i W represents the basic weights. d，i W represents the dynamic weights, i.e., the real-time anti-interference weights output during the iteration phase. r,i This represents the risk weight, specifically the risk enhancement weight W when the threshold is exceeded. r，i ;

[0031] After normalization to eliminate the influence of weighted sum fluctuations, the final output is the weighted average temperature T. output The global temperature field state of the container is reflected by the following formula:

[0032]

[0033] In some alternative implementations, step b includes the following specific details:

[0034] The local main controller inside the refrigerated container communicates with the redundant system terminal via a protocol. The status of the local main controller is monitored in real time through both physical signals and protocol communication channels, triggering a three-level alarm, specifically:

[0035] Step b1, Hardware Monitoring:

[0036] Monitor the electrical signal characteristics of the physical transmission medium, monitor different signals according to different protocols, and for the RS-485 bus communication interface, continuously collect the differential voltage value V. diff =|V A -V B |;For CAN bus monitoring, recessive level voltage V recessive If the monitored value exceeds the threshold for three consecutive measurement cycles, the hardware connection is deemed to have failed. The failure determination formula is as follows:

[0037]

[0038] Among them, V diff This represents the differential signal voltage. The industry standard minimum effective signal threshold is 0.2V. A As the positive signal, V B For negative signals, V cc dt represents the bus power supply voltage, and dt represents the fault duration determination time.

[0039] Step b2, Software Monitoring:

[0040] Step b 21 Dynamic heart rate monitoring:

[0041] Reference period T base =2×τ n , τ n Indicates the nominal response time reference;

[0042] Index retreats and retests: T k =T base ×(1.5) k-1 k = 1, 2, 3;

[0043] Failure determination: Three consecutive no-response events with a total time T total ≥5Tbase;

[0044] Step b 22 Data integrity failure verification is divided into three levels: CRC check, protocol feature check, and semantic range check. CRC check is a commonly used verification standard. Protocol feature check verifies whether the frame structure conforms to requirements for proprietary protocols. Semantic range check determines whether the temperature exceeds the maximum adjustment range. The specific range is determined based on the target container, and the basic formula is:

[0045]

[0046] If the data exceeds this range, the business data is considered abnormal.

[0047] Step b3, Alarm and Switching Strategy:

[0048] Level 1 warning: If a single heartbeat is lost or a single-level data verification fails, a local audible and visual alarm will be activated and the log will be recorded.

[0049] Level 2 fault: Physical connection interruption or data integrity failure, triggering remote control console alarm, while local backup controller goes into standby mode;

[0050] Level 3 failure: Dual-mode failure or complete unresponsiveness of the local master controller. Immediately switch to the redundant control system and notify the engineer.

[0051] In some alternative implementations, step c includes the following specific content: the redundant control system acquires temperature-weighted fusion data from the controller and the self-deployed temperature sensors, calculates the temperature difference, and adjusts the time.

[0052] In some optional implementations, step d involves establishing a dynamic scoring model by real-time acquisition and analysis of the performance parameters of each control unit, thereby achieving optimal allocation of control authority. Specifically:

[0053] Step d1: The control performance multi-dimensional parameter acquisition and analysis system monitors and records in real time the key performance indicators of the local main / backup controller and redundant system during the temperature regulation process, including: dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy.

[0054] Step d2: Establish an environment-adaptive dynamic weighted scoring model:

[0055]

[0056] Among them, S is the comprehensive score, which combines all parameters to evaluate the adaptive dynamic response of the system; w1 represents the response time weight, w2 represents the adjustment rate weight, w3 represents the overshoot weight, w4 represents the steady-state accuracy weight, and t... resp Represents response time, the time required for the system output to stabilize after a change in the input signal; v adj This represents the actual adjustment amount, reflecting the value during the actual adjustment process of the system; v max represents the maximum value of the adjustment amount, and represents the maximum limit that the adjustment amount can reach; tanh(2σ) is a function used to adjust the smoothness of the model; ∈ is a constant used to adjust the sensitivity of the system and avoid the calculation instability caused by the value being too small;

[0057] The weighting coefficients in the scoring model are dynamically adjusted based on real-time operating conditions.

[0058] Step d3, Control Decision and Smooth Transition Mechanism: Establish a three-level decision-making logic to achieve optimal allocation of control:

[0059] Step d 31 Real-time performance comparison: Calculates the comprehensive score S of each control unit every second. 主控 S 备用 S 冗余 ;

[0060] Step d 32 Takeover trigger conditions:

[0061] Emergency Takeover: When S 当前 <0.7×S 最优 And it lasts for 120 seconds;

[0062] Advantageous takeover: When S 冗余 >1.15×max(S 主控 S 备用 ).

[0063] In some alternative implementations, step e specifically includes the following:

[0064] Step e1: Multi-source data fusion and feature extraction, receiving real-time fused temperature data T from each container. 融合 Calculate the current temperature error e = T set -T 融合 Both and their rate of change, ec, are quantized to the fuzzy universe of discourse, and their membership degrees in the fuzzy subset are calculated using triangular membership functions, as follows:

[0065] Acquire the temperature values ​​of each sensor, denoted as T1, T2, ..., T n The data from each sensor is assigned a weight f1, f2, ..., f based on its location or accuracy. n The temperature data is weighted according to weights, and the calculation formula is as follows:

[0066]

[0067] Among them, T avg This represents the weighted average of the current temperatures, where I represents the weighting adjustment factor, which is a constant.

[0068] The formula for calculating the membership function μ(x) of a triangle is as follows:

[0069]

[0070] Where a is the left endpoint of the triangle, b is the peak point of the triangle, and c is the right endpoint of the triangle. In fuzzy control, given an input value x, the membership degree of the input value in different fuzzy subsets can be calculated based on the triangle membership function;

[0071] For temperature error e, assume that its fuzzy set includes three categories: low, medium and high. Each category corresponds to a triangular membership function with different endpoints a1, b1, c1, a2, b2, c2, a3, b3, c3. Based on the specific value of e, calculate the membership degree of the corresponding category.

[0072] Step e2: The adaptive fuzzy inference engine, based on a rule base trained using historical operating data, maps (e, ec) to the PID parameter adjustment amount (ΔK) through fuzzy inference. p ,ΔK i ,ΔK d ), K p K is the proportional gain, used to control the relationship between the system response and temperature error, reducing instantaneous error. i K is the integral gain, used to eliminate steady-state error. The control input is adjusted by accumulating the error. d The differential gain is used to suppress overshoot and oscillations in the system. The response speed is controlled by calculating the rate of change of the error. The rule design follows the principle of "prioritizing the elimination of large errors and suppressing oscillations for small errors," and uses the centroid method to defuzzify and output the precise value ΔK. p ,ΔK i ,ΔK d The formula for the center of gravity method is as follows:

[0073]

[0074] Where C is a weighting constant, and the output value x represents the position; for each PID parameter K p K i K d Calculate the centroid position of its fuzzy output to obtain the precise value ΔK for each parameter. p ,ΔK i ,ΔK d ;

[0075] Step e3: Update the PID parameters online based on the real-time inference results:

[0076] K p ′=K p +η p ·ΔK p

[0077] K i ′=K i +η i ·ΔK i

[0078] K d ′=K d +η d ·ΔK d

[0079] Where, η p η i η d The adaptive learning rate coefficient is dynamically adjusted based on the historical response characteristics of the container; the updated parameters are substituted into the PID formula to generate the control quantity formula as follows:

[0080]

[0081] In some alternative implementations, the fuzzy universe of discourse described in step e1 is [-6, 6].

[0082] A redundant control system for a ship refrigerated container cluster is characterized by comprising a local control layer, an edge processing layer, a wireless transmission layer, and a redundant control layer. The local control layer includes a local main controller, a local backup controller, and a variable frequency compressor. The local main controller and the local backup controller communicate with a protocol board through a private protocol. The protocol board transmits the status information of the local main controller and the local backup controller to the edge processing layer.

[0083] The edge processing layer includes multiple temperature sensors, an efficiency comparison module, a decision module, an output module, and wireless terminals. The multiple temperature sensors are deployed in various locations inside the refrigerated container, and the external communication interface is an RS485 bus. The efficiency comparison module compares the control effects of the local main controller, the backup controller, and the remote redundant control module based on the detection results of the protocol board, and then outputs the comparison results to the decision module. The decision module judges and controls the controller of the refrigerated container based on the efficiency comparison results. The output module outputs PWM to the variable frequency compressor of the refrigerated container to control the power level. The wireless terminal consists of multiple sensors, each configured to transmit and receive data.

[0084] The redundant control layer includes a historical database, an adaptive algorithm, and fuzzy PID control. The historical database stores data fused from multiple temperature sensors for preprocessing. The subsequent adaptive fuzzy PID control retrieves data from the historical database and deploys an adaptive algorithm for optimization and adjustment.

[0085] The wireless transmission layer includes a wireless gateway that receives data from all refrigerated containers on the ship. Part of the data is uploaded to a cloud database for fault comparison and performance comparison, while another part is sent to the historical database of the redundant control system. The data from the redundant control system is then distributed to the output modules of all refrigerated containers on the ship via the wireless gateway.

[0086] In some optional implementations, the status information of the local master controller and the local backup controller includes temperature output value, temperature setpoint, control timestamp, output PWM duty cycle, local controller model, and local backup controller model.

[0087] In some optional implementations, the control performance comparison parameters of the local main controller, the backup controller, and the remote redundant control module are dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy.

[0088] The beneficial effects of this application are as follows: The redundant control method and system for refrigerated container clusters provided in this application effectively eliminate the impact of single-point failures through the strategic deployment of multiple temperature sensors combined with a triple dynamic weight allocation mechanism, ensuring continuous and stable temperature output values. When any sensor fails, the system can still maintain a certain measurement accuracy, completely avoiding the risk of significant temperature data misalignment. By adopting local primary / backup controller fault detection and performance comparison, a third redundant control system is introduced to reduce the difference in refrigeration effects between refrigerated containers, ensuring that refrigerated containers always have a good refrigeration effect. For ship-wide fault takeover, a short-time reliable takeover mechanism is also realized, avoiding the risk of cargo quality damage due to the failure of local primary / backup controllers. At the same time, this system and method are not targeted at any specific ship and are applicable to various container ships. Attached Figure Description

[0089] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0090] Figure 1 This is a block diagram of the redundant control system for ship refrigerated container clusters in the embodiments of this application;

[0091] Figure 2 This is an overall architecture diagram of the redundancy control method for ship refrigerated container clusters in the embodiments of this application;

[0092] Figure 3 This is a logic diagram for temperature data fusion in an embodiment of this application;

[0093] Figure 4 This is a flowchart illustrating the control decision-making process in an embodiment of this application.

[0094] Figure 5 This is the adaptive fuzzy PID optimization diagram in the embodiments of this application. Detailed Implementation

[0095] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0096] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0097] It should be understood that the sequence number of each step in the embodiment does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0098] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0099] The features and performance of this application will be further described in detail below with reference to the embodiments.

[0100] Refrigerated containers heavily rely on local master controllers for temperature control. Some container brands use simple algorithms and lack sufficient precision in their controllers, making it difficult to meet the temperature control requirements of cargo. Furthermore, when the local master controller fails, refrigerated containers commonly suffer from missing backups or inadequate backup controller performance, hindering the maintenance of continuous temperature control. At the wireless control level, current technology is limited by the performance and reliability of local controllers. If a local controller malfunctions, it can lead to unstable execution of commands from the remote control console, poor fault tolerance, and the risk of system paralysis. Simultaneously, current container control methods are mostly limited to single, discrete management, failing to achieve clustered collaborative control and providing a consistent level of control capability for different vessels and multi-brand mixed loading scenarios, resulting in low control efficiency and poor system portability.

[0101] This application provides a redundant control method and system for ship refrigerated container clusters, which solves the problem of uncontrolled temperature inside the container caused by failure of the local or local backup controller.

[0102] like Figure 1As shown, a redundant control system for a ship's refrigerated container cluster includes a local control layer, an edge processing layer, a wireless transmission layer, and a redundant control layer.

[0103] 1. The local control layer includes a local main controller, a local backup controller, and a variable frequency compressor. The local main controller and local backup controller were designed by the original refrigerated container manufacturer. This system focuses on fault detection of the local main controller / local backup controller and comparison of control effects with the redundant control system, without modifying the original controller circuitry. The protocol board primarily communicates with the local controller of the refrigerated container via a proprietary protocol. This proprietary protocol will be further negotiated with various refrigerated container companies. It receives information such as the local controller's temperature output value, temperature setpoint, control timestamp, output PWM duty cycle, local controller model, and local backup controller model, and transmits this information to the edge processing layer. Another important function of the protocol board is fault detection, detecting whether the local main controller / local backup controller has failed and transmitting the controller status to the edge processing layer in real time for performance comparison.

[0104] 2. The edge processing layer mainly includes four temperature sensors, a performance comparison module, a decision-making module, an output module, and a wireless terminal. The four temperature sensors are deployed inside the refrigerated container, specifically PT100 temperature sensors. It should be noted that these four temperature sensors are not shared with the existing refrigerated container control system; they are dedicated redundant control devices. These four temperature sensors are deployed in four locations within the refrigerated container: the front top air supply area, the central core area, the bottom of the rear door, and the side return air area. The external communication interface is an RS485 bus.

[0105] 2.1. The performance comparison module compares the control performance of the local primary / backup controller and the remote redundant control module based on the protocol board test results. This comparison primarily focuses on dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy. The comparison results are then output to the decision module. If the local primary controller is functioning normally, the comparison is made between the local primary controller and the remote redundant control module. If the local primary controller is faulty, the comparison is made between the local backup controller and the remote redundant control module.

[0106] 2.2. The decision-making module determines which controller should operate the refrigerated container based on performance comparison results. If the local main controller performs better, the redundant control module only performs monitoring and transmits the monitoring data to the wireless gateway. Otherwise, the redundant control module will take over control of the refrigerated container from the local main controller. Similarly, if the local backup controller performs better, the redundant control module also only performs monitoring and transmits the monitoring data to the wireless gateway. Otherwise, the redundant control module will take over control of the refrigerated container from the local backup controller.

[0107] 2.3. Output module, mainly for outputting PWM for the variable frequency compressor of refrigerated container to control the power.

[0108] 2.4. Wireless Terminal: Each refrigerated container is equipped with one, primarily responsible for data transmission and reception. It sends data from the refrigerated container to the wireless gateway and simultaneously receives data from the wireless gateway to control the refrigerated container. This wireless terminal can specifically employ wireless technologies such as LoRa / Zigbee.

[0109] 3. The wireless transmission layer includes a wireless gateway, which receives data from all refrigerated containers on the ship. Part of the data is uploaded to the cloud database for fault comparison and performance comparison, while the other part is sent to the historical database of the redundant control system. The data from the redundant control system is then sent to the output modules of all refrigerated containers on the ship through the wireless gateway.

[0110] 4. The redundant control layer includes a historical database, an adaptive algorithm, and fuzzy PID control. The historical database stores data fused from multiple temperature sensors for preprocessing. The subsequent adaptive fuzzy PID control retrieves data from the historical database and deploys an adaptive algorithm for optimization and adjustment.

[0111] like Figure 2 As shown, the core of this solution lies in:

[0112] Step a: Deploy multiple temperature sensors inside each refrigerated container and use multi-source data fusion technology to obtain ambient temperature information;

[0113] Step b: Monitor the operating status of the local main controller of the refrigerated container in real time, determine whether it has failed, and issue an alarm signal;

[0114] Step c: The redundant system monitors and stores all key temperature data of refrigerated containers in real time, and performs simple preprocessing to form a historical temperature database.

[0115] Step d: Compare the control performance of the local primary or backup controller with that of the redundant system, and decide on the output signal weight;

[0116] Step e: Utilize the accumulated historical database to dynamically adjust and optimize the adaptive fuzzy PID control rules within the redundant system;

[0117] Step f: Collect and statistically analyze the local main / backup controller models, controller status, and performance comparison results, and upload them to the cloud to build a device performance knowledge base.

[0118] Furthermore, multiple temperature sensors are deployed within each refrigerated container, and multi-source data fusion technology is used to obtain highly reliable environmental status information. Specific steps are as follows: Figure 3 As shown.

[0119] Deploy n temperature sensors in a refrigerated container and output the global temperature field status through a triple-weighted fusion mechanism:

[0120] Step a1: Calculate the initial arithmetic mean temperature using the following formula.

[0121]

[0122] Among them, T i This indicates the measured temperature value of each temperature sensor.

[0123] Step a2, based on temperature deviation and rate of change Calculate dynamic weights

[0124]

[0125] in, Let represent the dynamic weights of the sensor in the k-th iteration. K represents the weighted average temperature calculated from sensor data at the k-th iteration. i K represents the position sensitivity coefficient. i ∈[0.6, 1.2], A i A represents the temperature change suppression coefficient. i ∈[0.3, 0.8].

[0126] Average value updated based on dynamic weights

[0127]

[0128] And through convergence conditions Continue until convergence, or until the number of iterations k≥3 controls the computational precision.

[0129] Step a3: When the temperature exceeds the preset threshold, the segmented adjustment of the risk weight W is activated. r，i :

[0130]

[0131] Where, Θ high Indicates the high temperature threshold, Θ low This indicates the low temperature threshold.

[0132] When T i When the temperature is above 5℃, the square of the weight increases, amplifying the risk of overheating; when T i At temperatures below 2°C, the weighted power function decreases, mitigating the risk of overcooling.

[0133] Step a4, Triple Fusion Output:

[0134] W final，i =0.5·W s，i +0.3·W d，i +0.2·W r，i

[0135] Among them, W s，i W represents the basic weights. d，i W represents the dynamic weights, i.e., the real-time anti-interference weights output during the iteration phase. r,i This represents the risk weight, specifically the risk enhancement weight W when the threshold is exceeded. r，i .

[0136] After normalization to eliminate the influence of weighted sum fluctuations, the final output is the weighted average temperature T. output The global temperature field state of the container is reflected by the following formula:

[0137]

[0138] Furthermore, the local main controller inside the refrigerated container communicates with the redundant system terminal boards via protocols, including but not limited to mainstream communication methods such as serial port, CAN, IIC, SPI, and ModueBus. Specific protocols need to be adjusted based on the products of each container brand, communicating with the local main controller of the refrigerated container according to the proprietary protocols of each brand. For monitoring the local controller, both hardware and software monitoring are employed.

[0139] Step b includes the following specific details:

[0140] The local main controller inside the refrigerated container communicates with the redundant system terminal via a protocol. It monitors the status of the local main controller in real time through both physical signals and protocol communication channels, triggering a three-level alarm, specifically:

[0141] Step b1, Hardware Monitoring:

[0142] Monitor the electrical signal characteristics of the physical transmission medium, monitor different signals according to different protocols, and for the RS-485 bus communication interface, continuously collect the differential voltage value V. diff =|V A -V B |;For CAN bus monitoring, recessive level voltage V recessive If the monitored value exceeds the threshold for three consecutive measurement cycles, the hardware connection is deemed to have failed. The failure determination formula is as follows:

[0143]

[0144] Among them, V diff This represents the differential signal voltage. The industry standard minimum effective signal threshold is 0.2V. AAs the positive signal, V B For negative signals, V cc dt represents the bus power supply voltage, and dt represents the fault duration determination time.

[0145] Step b2, Software Monitoring:

[0146] Step b 21 Dynamic heart rate monitoring:

[0147] Reference period T base =2×τ n , τ n Indicates the nominal response time reference;

[0148] Index retreats and retests: T k =T base ×(1.5) k-1 k = 1, 2, 3;

[0149] Failure determination: Three consecutive no-response events with a total time T total ≥5Tbase.

[0150] Step b 22 Data integrity failure verification is divided into three levels: CRC check, protocol feature check, and semantic range check. CRC check is a commonly used verification standard. Protocol feature check verifies whether the frame structure conforms to requirements for proprietary protocols. Semantic range check determines whether the temperature exceeds the maximum adjustment range. The specific range is determined based on the target container, and the basic formula is:

[0151]

[0152] If the data exceeds this range, the business data is considered abnormal.

[0153] Step b3, Alarm and Switching Strategy:

[0154] Level 1 warning: If a single heartbeat is lost or a single-level data verification fails, a local audible and visual alarm will be activated and the log will be recorded.

[0155] Level 2 fault: Physical connection interruption or data integrity failure, triggering remote control console alarm, while local backup controller goes into standby mode.

[0156] Level 3 failure: Dual-mode failure or complete unresponsiveness of the local master controller. Immediately switch to the redundant control system and notify the engineer.

[0157] Furthermore, the redundant control system collects temperature-weighted fusion data from the controller and self-deployed temperature sensors to calculate the temperature difference and adjustment time.

[0158] Furthermore, in step d, by collecting and analyzing the performance parameters of each control unit in real time, a dynamic scoring model is established to achieve the optimal allocation of control authority, such as... Figure 4 As shown, specifically:

[0159] Step d1: The control performance multi-dimensional parameter acquisition and analysis system monitors and records in real time the key performance indicators of the local main / backup controller and redundant system during the temperature regulation process, including: dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy.

[0160] Step d2: Establish an environment-adaptive dynamic weighted scoring model:

[0161]

[0162] Among them, the S-comprehensive score evaluates the system's adaptive dynamic response by combining all parameters; w1 represents the response time weight, w2 represents the settling rate weight, w3 represents the overshoot weight, and w4 represents the overshoot weight.

[0163] Represents the steady-state accuracy weight, t resp Represents response time, the time required for the system output to stabilize after a change in the input signal; v adj This represents the actual adjustment amount, reflecting the value during the actual adjustment process of the system; v max represents the maximum value of the adjustment amount, and represents the maximum limit that the adjustment amount can reach; tanh(2σ) is a function used to adjust the smoothness of the model; ∈ is a constant used to adjust the sensitivity of the system and avoid the calculation instability caused by the value being too small.

[0164] The weight coefficients in the scoring model are dynamically adjusted according to real-time operating conditions.

[0165] Step d3, Control Decision and Smooth Transition Mechanism: Establish a three-level decision-making logic to achieve optimal allocation of control:

[0166] Step d 31 Real-time performance comparison: Calculates the comprehensive score S of each control unit every second. 主控 S 备用 S 冗余 ;

[0167] Step d 32 Takeover trigger conditions:

[0168] Emergency Takeover: When S 当前 <0.7×S 最优 And it lasts for 120 seconds;

[0169] Advantageous takeover: When S 冗余 >1.15×max(S) 主控 S 备用 ).

[0170] like Figure 5 As shown, the refrigerated container system internally deploys fuzzy PID and adaptive algorithms. The redundant control system uses fuzzy PID for control and adaptively adjusts according to different containers. Step e specifically includes the following:

[0171] Step e1: Multi-source data fusion and feature extraction, receiving real-time fused temperature data T from each container. 融合 Calculate the current temperature error e = T set -T 融合 And its rate of change ec, both are quantized to the fuzzy universe of discourse (typically in the range [-6, 6]), and their membership degree in the fuzzy subset is calculated using the triangular membership function, as follows:

[0172] Acquire the temperature values ​​of each sensor, denoted as T1, T2, ..., T n The data from each sensor is assigned a weight f1, f2, ..., f based on its location or accuracy. n The temperature data is weighted according to weights, and the calculation formula is as follows:

[0173]

[0174] Among them, T avg This represents the weighted average of the current temperature. I represents the weight adjustment factor, which is a constant. Its function is to prevent the weight from becoming infinitely large when the denominator is too small or zero. By introducing the constant I, the overall magnitude of the weight can be controlled to ensure the stability of the weighted average value.

[0175] The formula for calculating the membership function μ(x) of a triangle is as follows:

[0176]

[0177] Where a is the left endpoint of the triangle, b is the peak point of the triangle, and c is the right endpoint of the triangle. In fuzzy control, given an input value x, the membership degree of the input value in different fuzzy subsets can be calculated based on the triangle membership function.

[0178] For the temperature error e, assume that its fuzzy set includes three categories: low, medium and high. Each category corresponds to a triangular membership function with different endpoints a1, b1, c1, a2, b2, c2, a3, b3, c3. Based on the specific value of e, the membership degree of the corresponding category is calculated.

[0179] Step e2: The adaptive fuzzy inference engine, based on a rule base trained using historical operating data, maps (e, ec) to the PID parameter adjustment amount (ΔK) through fuzzy inference. p ΔK iΔK d ), K p K is the proportional gain, used to control the relationship between the system response and temperature error, reducing instantaneous error. i K is the integral gain, used to eliminate steady-state error. The control input is adjusted by accumulating the error. d The differential gain is used to suppress overshoot and oscillations in the system. The response speed is controlled by calculating the rate of change of the error. The rule design follows the principle of "prioritizing the elimination of large errors and suppressing oscillations for small errors," and uses the centroid method to defuzzify and output the precise value ΔK. p ΔK i ΔK d The formula for the center of gravity method is as follows:

[0180]

[0181] Where C is a weighting constant used to optimize the tuning of PID parameters, and the output value x represents the position for each PID parameter K. p K i K d Calculate the centroid position of its fuzzy output to obtain the precise value ΔK for each parameter. p ΔK i ΔK d .

[0182] Step e3: Update the PID parameters online based on the real-time inference results:

[0183] K p ′=K p +η p ·ΔK p

[0184] K i ′=K i +η i ·ΔK i

[0185] K d ′=K d +η d ·ΔK d

[0186] Where, η p η i η d The adaptive learning rate coefficient is dynamically adjusted based on the historical response characteristics of the container; the updated parameters are substituted into the PID formula to generate the control quantity formula as follows:

[0187]

[0188] Example 1

[0189] In this embodiment, four temperature sensors are deployed inside the refrigerated container, specifically PT100 temperature sensors. It should be noted that these four temperature sensors are not shared with the original refrigerated container control system; they are dedicated redundant control devices. These four temperature sensors are deployed in four locations within the refrigerated container: the front top air supply area, the central core area, the bottom of the rear door, and the side return air area. The external communication interface is an RS485 bus.

Claims

1. A method for redundancy control of refrigerated container clusters on ships, characterized in that, Includes the following steps: Step a: Deploy multiple temperature sensors inside each refrigerated container and use multi-source data fusion technology to obtain ambient temperature information; Step b: Monitor the operating status of the local main controller of the refrigerated container in real time, determine whether it has failed, and issue an alarm signal; Step c: The redundant system monitors and stores all key temperature data of refrigerated containers in real time, and performs simple preprocessing to form a historical temperature database. Step d: Compare the control performance of the local primary or backup controller with that of the redundant system, and decide on the output signal weight; Step e: Utilize the accumulated historical database to dynamically adjust and optimize the adaptive fuzzy PID control rules within the redundant system; Step f: Collect and statistically analyze the local main / backup controller models, controller status, and performance comparison results, and upload them to the cloud to build a device performance knowledge base.

2. The redundancy control method for a ship's refrigerated container cluster according to claim 1, characterized in that, Step a includes the following specific details: Deploy n temperature sensors in a refrigerated container and output the global temperature field status through a triple-weighted fusion mechanism: Step a1: Calculate the initial arithmetic mean temperature using the following formula. Among them, T i This indicates the measured temperature value of each temperature sensor. Step a2, based on temperature deviation and rate of change Calculate dynamic weights in, Let represent the dynamic weights of the sensor in the k-th iteration. K represents the weighted average temperature calculated from sensor data at the k-th iteration. i K represents the position sensitivity coefficient. i ∈[0.6, 1.2], A i A represents the temperature change suppression coefficient. i ∈[0.3, 0.8], Average value updated based on dynamic weights And through convergence conditions Continue until convergence, or until the number of iterations k≥3 controls the computational precision; Step a3: When the temperature exceeds the preset threshold, the segmented adjustment of the risk weight W is activated. r，i : Where, Θ high Indicates the high temperature threshold, Θ low Indicates the low temperature threshold; When T i When the temperature is above 5℃, the square of the weight increases, amplifying the risk of overheating; when T i At temperatures below 2℃, the weighted power function decreases, mitigating the risk of overcooling. Step a4, Triple Fusion Output: IN final，i =0.5 W s，i +0.3 W d，i +0.2 W r，i Among them, W s，i W represents the basic weights. d，i W represents the dynamic weights, i.e., the real-time anti-interference weights output during the iteration phase. r，i This represents the risk weight, specifically the risk enhancement weight W when the threshold is exceeded. r，i ; After normalization to eliminate the influence of weighted sum fluctuations, the final output is the weighted average temperature T. output The global temperature field state of the container is reflected by the following formula:

3. The redundancy control method for a ship's refrigerated container cluster according to claim 2, characterized in that, Step b includes the following specific details: The local main controller inside the refrigerated container communicates with the redundant system terminal via a protocol. The status of the local main controller is monitored in real time through both physical signals and protocol communication channels, triggering a three-level alarm, specifically: Step b1, Hardware Monitoring: Monitor the electrical signal characteristics of the physical transmission medium, monitor different signals according to different protocols, and for the RS-485 bus communication interface, continuously collect the differential voltage value V. diff =|V A -V B |;For CAN bus monitoring, recessive level voltage V recessive If the monitored value exceeds the threshold for three consecutive measurement cycles, the hardware connection is determined to be faulty. The fault determination formula is as follows: Among them, V diff This represents the differential signal voltage. The industry standard minimum effective signal threshold is 0.2V. A As the positive signal, V B For negative signals, V cc dt represents the bus power supply voltage, and dt represents the fault duration determination time. Step b2, Software Monitoring: Step b 21 Dynamic heart rate monitoring: Reference period T base =2×τ n , τ n Indicates the nominal response time reference; Index retreats and retests: T k =T base ×(1.5) k-1 k = 1, 2, 3; Failure determination: Three consecutive no-response events with a total time T total ≥5Tbase; Step b 22 Data integrity failure verification is divided into three levels: CRC check, protocol feature check, and semantic range check. CRC check is a commonly used verification standard. Protocol feature check verifies whether the frame structure conforms to requirements for proprietary protocols. Semantic range check determines whether the temperature exceeds the maximum adjustment range. The specific range is determined based on the target container, and the basic formula is: If the data exceeds this range, the business data is considered abnormal. Step b3, Alarm and Switching Strategy: Level 1 warning: If a single heartbeat is lost or a single-level data verification fails, a local audible and visual alarm will be activated and the log will be recorded. Level 2 fault: Physical connection interruption or data integrity failure, triggering remote control console alarm, while local backup controller goes into standby mode; Level 3 failure: Dual-mode failure or complete unresponsiveness of the local master controller. Immediately switch to the redundant control system and notify the engineer.

4. The redundancy control method for a ship's refrigerated container cluster according to claim 3, characterized in that, Step c includes the following specific content: the redundant control system acquires temperature weighted fusion data from the controller and the self-deployed temperature sensor, calculates the temperature difference and adjustment time.

5. The redundancy control method for a ship's refrigerated container cluster according to claim 4, characterized in that, In step d, by collecting and analyzing the performance parameters of each control unit in real time, a dynamic scoring model is established to achieve the optimal allocation of control authority. Specifically: Step d1: The control performance multi-dimensional parameter acquisition and analysis system monitors and records in real time the key performance indicators of the local main / backup controller and redundant system during the temperature regulation process, including: dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy. Step d2: Establish an environment-adaptive dynamic weighted scoring model: Among them, S is the comprehensive score, which combines all parameters to evaluate the adaptive dynamic response of the system; w1 represents the response time weight, w2 represents the adjustment rate weight, w3 represents the overshoot weight, w4 represents the steady-state accuracy weight, and t... resp Represents response time, the time required for the system output to stabilize after a change in the input signal; v adj This represents the actual adjustment amount, reflecting the value during the actual adjustment process of the system; v max represents the maximum value of the adjustment amount, and represents the maximum limit that the adjustment amount can reach; tanh(2σ) is a function used to adjust the smoothness of the model; ∈ is a constant used to adjust the sensitivity of the system and avoid the calculation instability caused by the value being too small; The weighting coefficients in the scoring model are dynamically adjusted based on real-time operating conditions. Step d3, Control Decision and Smooth Transition Mechanism: Establish a three-level decision-making logic to achieve optimal allocation of control: Step d 31 Real-time performance comparison: Calculates the comprehensive score S of each control unit every second. 主控 S 备用 S 冗余 ; Step d 32 Takeover trigger conditions: Emergency Takeover: When S 当前 <0.7×S 最优 And it lasts for 120 seconds; Advantageous takeover: When S 冗余 >1.15×max(S 主控 S 备用 ).

6. The redundancy control method for a ship's refrigerated container cluster according to claim 5, characterized in that, Step e specifically includes the following: Step e1: Multi-source data fusion and feature extraction, receiving real-time fused temperature data T from each container. 融合 Calculate the current temperature error e = T set -T 融合 Both and their rate of change, ec, are quantized to the fuzzy universe of discourse, and their membership degrees in the fuzzy subset are calculated using triangular membership functions, as follows: Acquire the temperature values ​​of each sensor, denoted as T1, T2, ..., T n The data from each sensor is assigned a weight f1, f2, ..., f based on its location or accuracy. n The temperature data is weighted according to weights, and the calculation formula is as follows: Among them, T avg This represents the weighted average of the current temperatures, where I represents the weighting adjustment factor, which is a constant. The formula for calculating the membership function μ(x) of a triangle is as follows: Where a is the left endpoint of the triangle, b is the peak point of the triangle, and c is the right endpoint of the triangle. In fuzzy control, given an input value x, the membership degree of the input value in different fuzzy subsets can be calculated based on the triangle membership function; For temperature error e, assume that its fuzzy set includes three categories: low, medium and high. Each category corresponds to a triangular membership function with different endpoints a1, b1, c1, a2, b2, c2, a3, b3, c3. Based on the specific value of e, calculate the membership degree of the corresponding category. Step e2: The adaptive fuzzy inference engine, based on a rule base trained using historical operating data, maps (e, ec) to the PID parameter adjustment amount (ΔK) through fuzzy inference. p ΔK i ΔK d ), K p K is the proportional gain, used to control the relationship between the system response and temperature error, reducing instantaneous error. i K is the integral gain, used to eliminate steady-state error. The control input is adjusted by accumulating the error. d The differential gain is used to suppress overshoot and oscillations in the system. The response speed is controlled by calculating the rate of change of the error. The rule design follows the principle of "prioritizing the elimination of large errors and suppressing oscillations of small errors," and uses the centroid method to defuzzify and output the precise value ΔK. p ΔK i ΔK d The formula for the center of gravity method is as follows: Where C is a weighting constant, and the output value x represents the position; for each PID parameter K p K i K d Calculate the centroid position of its fuzzy output to obtain the precise value ΔK for each parameter. p ΔK i ΔK d ; Step e3: Update the PID parameters online based on the real-time inference results: K p ′=K p +η p ·ΔK p K i ′=K i +η i ·ΔK i K d ′=K d +η d ·ΔK d Where, η p ,η i ,η d The adaptive learning rate coefficient is dynamically adjusted based on the historical response characteristics of the container. Substituting the updated parameters into the PID formula, the control quantity formula is generated as follows:

7. The redundancy control method for a ship's refrigerated container cluster according to claim 6, characterized in that, The fuzzy universe of discourse mentioned in step e1 is [-6, 6].

8. A control system for a redundancy control method for a ship refrigerated container cluster as described in claim 6 or 7, characterized in that, It includes a local control layer, an edge processing layer, a wireless transmission layer, and a redundancy control layer. The local control layer includes a local main controller, a local backup controller, and a variable frequency compressor. The local main controller and the local backup controller communicate with a protocol board through a private protocol. The protocol board transmits the status information of the local main controller and the local backup controller to the edge processing layer. The edge processing layer includes multiple temperature sensors, an efficiency comparison module, a decision module, an output module, and wireless terminals. The multiple temperature sensors are deployed in various locations inside the refrigerated container, and the external communication interface is an RS485 bus. The efficiency comparison module compares the control effects of the local main controller, the backup controller, and the remote redundant control module based on the detection results of the protocol board, and then outputs the comparison results to the decision module. The decision module judges and controls the controller of the refrigerated container based on the efficiency comparison results. The output module outputs PWM to the variable frequency compressor of the refrigerated container to control the power level. The wireless terminal consists of multiple sensors, each configured to transmit and receive data. The redundant control layer includes a historical database, an adaptive algorithm, and fuzzy PID control. The historical database stores data fused from multiple temperature sensors for preprocessing. The subsequent adaptive fuzzy PID control retrieves data from the historical database and deploys an adaptive algorithm for optimization and adjustment. The wireless transmission layer includes a wireless gateway that receives data from all refrigerated containers on the ship. Part of the data is uploaded to a cloud database for fault comparison and performance comparison, while another part is sent to the historical database of the redundant control system. The data from the redundant control system is then distributed to the output modules of all refrigerated containers on the ship via the wireless gateway.

9. A redundant control system for a ship's refrigerated container cluster according to claim 8, characterized in that, The status information of the local master controller and the local backup controller includes temperature output value, temperature setpoint, control timestamp, output PWM duty cycle, local controller model, and local backup controller model.

10. A redundancy control method for a ship's refrigerated container cluster according to claim 8, characterized in that, The control performance comparison parameters of the local main controller, backup controller and remote redundant control module are dynamic response time, temperature regulation rate, overshoot, and steady-state control accuracy.

Images