A precise control method and system for intelligent control cabinet based on fusion fuzzy algorithm

Abstract

The present invention relates to the technical field of intelligent control cabinets, and in particular to a precise control method and system for intelligent control cabinets based on a fusion fuzzy algorithm. The method comprises acquiring the control data of the control cabinet, performing preprocessing based on the acquired control data, performing feature extraction based on the preprocessed control data, constructing a fuzzy control model using the extracted features as input, introducing a priority matrix and an energy efficiency optimization strategy, converting the control quantity obtained by defuzzification into a control signal of an actuator, and performing intelligent collaborative control of the actuator; performing model optimization and fault prediction based on the collaborative control results, and performing precise control of the control cabinet based on the prediction results. The present invention performs model optimization and fault prediction based on the collaborative control results, which enables the model to continuously adapt to changes in the system. At the same time, a real-time environmental compensation factor is introduced to dynamically adjust the weights of the fuzzy rules, further improving the adaptability of the control cabinet to environmental changes.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control cabinets, and in particular to a precise control method and system for an intelligent control cabinet based on a fusion fuzzy algorithm. Background Art

[0002] In many fields such as industrial automation, new energy, data centers, and intelligent buildings, control cabinets, as core control equipment, play a vital role in ensuring the stable operation of the system. Early control cabinet control mainly relied on simple manual operation. Operators adjusted the operating parameters of the equipment based on experience and actual observation. This method is not only inefficient, but also difficult to ensure the control accuracy and is easily affected by human factors. In order to improve the control accuracy and degree of automation of control cabinets, PID control algorithms are widely used in control cabinets. PID controllers perform proportional, integral, and differential operations on the system error and output corresponding control signals to adjust the operating state of the controlled object. The PID control algorithm has the advantages of simple structure, easy implementation, and strong robustness, and has achieved good control effects in many industrial process controls.

[0003] In practical applications, control cabinets need to simultaneously control multiple parameters, such as temperature, humidity, voltage, current, etc. There are often complex coupling relationships between these parameters. Changes in one parameter will affect the status of other parameters. Traditional PID control algorithms usually independently control a single parameter and have difficulty dealing with the coupling problem between multiple parameters, which can easily lead to poor system control effects or even instability. In addition, many actual controlled objects have nonlinear characteristics, and their input-output relationships are not simple linear relationships. Traditional PID control algorithms are designed based on linear system theory and have poor adaptability to nonlinear systems. When the operating point of the controlled object changes or is subject to external interference, the traditional PID controller may not be able to adjust the control parameters in time, resulting in a decrease in the system's control accuracy, or even oscillation or divergence. At this stage, there is a need for a precise control method and system for intelligent control cabinets based on a fusion fuzzy algorithm. Summary of the Invention

[0004] In order to solve the problems of poor adaptability in traditional control cabinet regulation and poor system regulation effect due to coupling between parameters, the present invention provides a precise regulation method and system for an intelligent control cabinet based on a fusion fuzzy algorithm.

[0005] In the first aspect, the present invention provides a precise control method for an intelligent control cabinet based on a fusion fuzzy algorithm, which adopts the following technical solutions:

[0006] A precise control method for an intelligent control cabinet based on a fusion fuzzy algorithm, comprising:

[0007] Obtain control data of the control cabinet, including environmental parameters and equipment operating status data inside the control cabinet;

[0008] Preprocessing is performed based on the acquired control data, including noise reduction using Kalman filtering and the introduction of an adaptive noise covariance adjustment mechanism;

[0009] Feature extraction is performed based on the pre-processed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates;

[0010] The extracted features are used as input to construct a fuzzy control model, including building a fuzzy rule base, a fuzzy reasoning mechanism and a defuzzification module;

[0011] Priority matrix and energy efficiency optimization strategy are introduced to convert the control quantity obtained by defuzzification into the control signal of the actuator, and the actuator is intelligently coordinated controlled;

[0012] Model optimization and fault prediction are performed based on the collaborative control results, and the control cabinet is precisely adjusted based on the prediction results.

[0013] Furthermore, the introduction of the adaptive noise covariance adjustment mechanism includes initializing the Kalman filter parameters according to the acquired control data, calculating the residual sequence of the data, and dynamically adjusting the process noise covariance and the measurement noise covariance using the variance of the residual sequence. The adjusted covariance is combined with the new measurement information to correct the state estimation. The dynamic adjustment formula is:

[0014] ,

[0015] ,

[0016] in, and are the adjustment coefficients for process noise covariance and measurement noise covariance, is the variance of the residual sequence, and are the process noise covariance and measurement noise covariance at the previous moment, respectively.

[0017] Furthermore, the feature extraction is performed based on the preprocessed control data, including calculating the temperature deviation and humidity deviation based on the preprocessed control data to extract the single time step deviation, then setting a sliding window of length L, the window containing L consecutive time step data, calculating the average voltage fluctuation rate within the window, and finally introducing an exponentially weighted average method to calculate the voltage fluctuation rate. The calculation formula of the exponentially weighted average voltage fluctuation rate is:

[0018] ,

[0019] in, For the forgetting factor, For the The exponentially weighted average voltage fluctuation rate in time steps, Expressed as voltage, Expressed as time difference.

[0020] Furthermore, the feature extraction based on the preprocessed control data also includes setting membership function parameters, fuzzifying the calculated deviation and fluctuation rate, introducing a density clustering algorithm to dynamically determine the number of fuzzy levels, calculating the triangular membership function, Gaussian membership function and trapezoidal membership function according to the fuzzy levels, mixing different membership functions by dynamically adjusting weights, and converting numerical values ​​into fuzzy linguistic variables.

[0021] Furthermore, the construction of the fuzzy rule base, the fuzzy reasoning mechanism and the defuzzification module includes taking the extracted features as input, completely traversing the input data set, counting the frequency of occurrence of each item and sorting them in descending order of frequency, using the FP-growth algorithm to mine association rules, recursively mining frequent item sets starting from the bottom of the FP-tree, generating association rules based on the frequent item sets, and screening out association rules with a confidence level greater than a threshold, and constructing a fuzzy rule base in the form of IF-THEN based on the screened association rules.

[0022] Furthermore, the construction of the fuzzy rule base, fuzzy reasoning mechanism and defuzzification module also includes using the Mamdani reasoning method to search for matching rules in the rule base based on the fuzzified input, calculating the matching degree of each matching rule, and then introducing evidence theory to correct the matching degree. The weighted average defuzzification method is used to calculate the control quantity based on the rule matching degree obtained by fuzzy reasoning and the central value of the output membership function.

[0023] Furthermore, the model optimization and fault prediction are performed according to the collaborative control results, including constructing a fitness function based on the control error and the control quantity change rate, encoding the association rules in the fuzzy rule base through binary coding, and selecting individuals with higher fitness from the current population as parents for genetic algorithm operations according to the value of the fitness function. Thereafter, a real-time environmental compensation factor is introduced, and the weights of the fuzzy rules are dynamically adjusted according to the environmental compensation factor to complete the model optimization.

[0024] The second aspect is a precise control system for intelligent control cabinets based on fusion fuzzy algorithms, including:

[0025] The data acquisition module is configured to: acquire control data of the control cabinet, including environmental parameters and equipment operating status data in the control cabinet;

[0026] The preprocessing module is configured to: perform preprocessing based on the acquired control data, including using Kalman filtering for noise reduction and introducing an adaptive noise covariance adjustment mechanism;

[0027] The feature extraction module is configured to: perform feature extraction based on the preprocessed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates;

[0028] The model module is configured to: construct a fuzzy control model using the extracted features as input, including constructing a fuzzy rule base, a fuzzy inference mechanism, and a defuzzification module;

[0029] The conversion module is configured to: introduce a priority matrix and an energy efficiency optimization strategy to convert the defuzzified control quantity into a control signal for the actuator, and the actuator performs intelligent collaborative control;

[0030] The output module is configured to perform model optimization and fault prediction based on the collaborative control results, and to precisely control the control cabinet based on the prediction results.

[0031] In a third aspect, the present invention provides a computer-readable storage medium storing a plurality of instructions, wherein the instructions are suitable for being loaded and executed by a processor of a terminal device, for a method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm.

[0032] In a fourth aspect, the present invention provides a terminal device comprising a processor and a computer-readable storage medium, wherein the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded and executed by the processor to implement the method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm.

[0033] In summary, the present invention has the following beneficial technical effects:

[0034] 1. The present invention calculates the residual sequence of the data and uses its variance to dynamically adjust the process noise covariance and measurement noise covariance, which can more accurately estimate the system state. When there is sudden noise interference in the environment where the control cabinet is located, the dynamically adjusted covariance can enable the Kalman filter to filter noise more effectively, improve the quality of the data, and thus enhance the accuracy of subsequent control.

[0035] 2. This invention not only extracts single-time-step deviations but also uses multi-time-scale analysis to extract fluctuation rates. Single-time-step deviations reflect the current state difference, while fluctuation rates reflect the parameter's changing trend over time. By calculating the average voltage fluctuation rate and the exponentially weighted average voltage fluctuation rate, a more comprehensive understanding of the voltage's dynamic changes can be achieved. This multi-dimensional feature extraction approach helps capture the complex dynamic characteristics of the system.

[0036] 3. The present invention uses the FP-growth algorithm to mine association rules, automatically discovers the association between input variables and output control quantities from a large amount of data, and constructs a fuzzy rule base. Compared with the traditional method of constructing a rule base based on expert experience, the present invention can more objectively and comprehensively mine the potential rules in the data and generate rules that are more in line with the actual system operation conditions.

[0037] 4. The present invention adopts the Mamdani reasoning method for fuzzy reasoning and introduces evidence theory to correct the matching degree. In the defuzzification stage, the weighted average defuzzification method is used to calculate the control quantity according to the rule matching degree and the central value of the output membership function. It can convert the fuzzy reasoning results into accurate control signals, providing an accurate basis for the coordinated control of the actuators.

[0038] 5. The present invention introduces a priority matrix and an energy efficiency optimization strategy to convert the control quantity obtained by defuzzification into a control signal of the actuator, thereby realizing intelligent collaborative control of the actuator. Through the priority matrix and the energy efficiency optimization strategy, the tasks of each actuator can be reasonably allocated according to the actual needs and operating status of the system, thereby realizing collaborative work among the actuators and improving the overall regulation efficiency and energy utilization efficiency of the system.

[0039] 6. The present invention performs model optimization and fault prediction based on the collaborative control results, constructs a fitness function based on the control error and the control quantity change rate, and optimizes the fuzzy rule base using a genetic algorithm, which enables the model to continuously adapt to changes in the system. At the same time, a real-time environmental compensation factor is introduced to dynamically adjust the weights of the fuzzy rules, further improving the adaptability of the control cabinet to environmental changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the overall process of a method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described in detail below with reference to the accompanying drawings.

[0042] Example 1

[0043] Reference Figure 1 , a precise control method of an intelligent control cabinet based on a fusion fuzzy algorithm in this embodiment includes:

[0044] Obtain control data of the control cabinet, including environmental parameters and equipment operating status data inside the control cabinet;

[0045] Preprocessing is performed based on the acquired control data, including noise reduction using Kalman filtering and the introduction of an adaptive noise covariance adjustment mechanism;

[0046] Feature extraction is performed based on the pre-processed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates;

[0047] The extracted features are used as input to construct a fuzzy control model, including building a fuzzy rule base, a fuzzy reasoning mechanism and a defuzzification module;

[0048] Priority matrix and energy efficiency optimization strategy are introduced to convert the control quantity obtained by defuzzification into the control signal of the actuator, and the actuator is intelligently coordinated controlled;

[0049] Model optimization and fault prediction are performed based on the collaborative control results, and the control cabinet is precisely adjusted based on the prediction results.

[0050] Specifically, a precise control method for an intelligent control cabinet based on a fusion fuzzy algorithm includes the following steps:

[0051] like Figure 1 As shown, S1, obtaining the control data of the control cabinet, including obtaining the environmental parameters and equipment operating status data in the control cabinet;

[0052] Various sensors are used to collect environmental parameters and equipment operating status information in the control cabinet in real time. The collection frequency is set to once every 100ms to ensure real-time data. In addition to traditional sensor data such as temperature, humidity, voltage, current, and pressure, a multimodal data fusion collection mechanism is introduced to collect vibration frequency and sound spectrum data of equipment in the control cabinet. These additional data can reflect the operating status of the equipment from different dimensions. For example, abnormal vibration frequency may indicate wear of mechanical parts of the equipment, and changes in the sound spectrum may indicate electrical component failure.

[0053] S2. Preprocessing the acquired control data, including noise reduction using Kalman filtering and introducing an adaptive noise covariance adjustment mechanism;

[0054] Based on the acquired data, Kalman filtering is used to reduce noise. First, the principle of Kalman filtering is explained. Kalman filtering is a recursive algorithm for estimating the state of a dynamic system. It combines the system's predicted information with new measurement information to obtain the optimal estimate of the system state. The algorithm is based on two core equations: the state equation and the measurement equation. The state equation is:

[0055] ,

[0056] in, represents the system state vector at time k, is the state transition matrix, which describes the evolution of the system state over time. is the system state vector at time k-1, is the control input matrix, is the control input vector at time k, is the process noise, which follows a Gaussian distribution , is the process noise covariance matrix, which reflects the statistical characteristics of the process noise.

[0057] The measurement equation is expressed as:

[0058] ,

[0059] in, is the measurement vector at time k, which is the data actually measured by the sensor. H is the measurement matrix, which establishes the connection between the system state and the measurement value. is the measurement noise, which follows a Gaussian distribution , is the measurement noise covariance matrix, which reflects the noise situation in the measurement process.

[0060] Next, the specific process of implementing Kalman filter noise reduction is described:

[0061] First, the state estimate and error covariance matrix are initialized, and prediction operations are performed at each time step k.

[0062] Estimated based on the state at the previous moment , use the state equation to predict the state estimate at the current moment, the formula is:

[0063] ,

[0064] in, is the state estimate of the previous moment, according to the error covariance matrix of the previous moment , calculate the prediction error covariance matrix at the current moment, the formula is:

[0065] ,

[0066] in, is the error covariance matrix of the previous moment, and at the current moment k, the measurement value is obtained from the sensor , combined with the predicted state estimate and prediction error covariance matrix obtained in the prediction step, and the measured value obtained in the measurement step, an update operation is performed to calculate the Kalman gain, the updated state estimate, and the updated error covariance respectively. The calculated Kalman gain is expressed as:

[0067] .

[0068] in, is the transposed matrix of the measurement matrix H, It is expressed as Kalman gain.

[0069] The updated state estimate is expressed as:

[0070] ,

[0071] in, is the optimal state estimate at time k, which is the best estimate of the true state of the system obtained by combining the predicted value and the measured value. is the predicted state estimate at time k, which is the state at time k predicted by the state equation based on the state estimate at time k-1. is the measurement value at time k.

[0072] The updated error covariance is expressed as:

[0073] ,

[0074] in, is the error covariance matrix after the k-th moment update, which represents the uncertainty of the state estimate after combining the measurement information at the k-th moment. It is the identity matrix, which is used to ensure the dimensional matching of matrix operations and to adjust the influence of Kalman gain on the prediction error covariance matrix.

[0075] In traditional Kalman filtering, the process noise covariance and the measurement noise covariance It is usually a fixed value. However, in actual application scenarios, the characteristics of noise often change with time and environment. Therefore, an adaptive noise covariance adjustment mechanism is introduced to dynamically adjust the noise covariance according to the real-time collected data. and At each time step k, the measurement residual is calculated The measurement residual represents the difference between the measured value and the measured value estimated based on the predicted state. The statistical analysis of the residual sequence is mainly to calculate the variance of the residual. The variance of the residual can reflect the degree of deviation between the measured value and the predicted value. When the variance increases, it means that the influence of the noise is increasing; when the variance decreases, it means that the influence of the noise is decreasing. According to the statistical characteristics of the residual, the process noise covariance is dynamically adjusted. and the measurement noise covariance The adjustment formula used is as follows:

[0076] ,

[0077] ,

[0078] in, and are the adjustment coefficients for process noise covariance and measurement noise covariance, is the variance of the residual sequence, and The process noise covariance and measurement noise covariance at the previous moment are respectively used. After completing the prediction and update steps of the Kalman filter and adjusting the noise covariance, the algorithm proceeds to the next time step k+1 and repeats all the above steps until all the control data has been processed. Through continuous iteration, the Kalman filter can always effectively reduce the noise of the data and adaptively adjust the noise covariance to adapt to complex and changing noise environments.

[0079] S3. Extract features based on the pre-processed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates;

[0080] Feature extraction is performed based on the pre-processed control data. The single time step deviation is the core feature that reflects the real-time status of the system. This is achieved by comparing the difference between the actual value and the set target value. The difference between the current temperature and the set target temperature is calculated to reflect the degree of deviation from the temperature control. For example, if the target temperature is 25°C and the current measured temperature is 26°C, the temperature deviation is +1°C, indicating that cooling control is required. Similarly, the difference between the measured humidity and the target humidity reflects the direction and magnitude of the humidity control deviation. The single time step voltage fluctuation rate can reflect the change in voltage between two adjacent time steps. Set The actual voltage at the kth time step, ) is the actual voltage at the k-1th time step, is the time step, then the voltage fluctuation rate in a single time step is The calculation formula is:

[0081] ,

[0082] In order to capture the long-term trend of voltage fluctuation, the sliding window technology is introduced to calculate the multi-time scale average voltage fluctuation rate. A sliding window with a length of L is set, and the window contains L consecutive time step data. For example, if L=10 and the time step is , then the time range covered by the window is 1s. For the kth time step, the average voltage fluctuation rate within the window is calculated , the formula is:

[0083] ,

[0084] In order to consider both the short-term impact and long-term trend of voltage fluctuation, the exponential weighted average (EWMA) method is introduced to calculate the voltage fluctuation rate. is the exponentially weighted average voltage fluctuation rate at the kth time step, and the calculation formula of the exponentially weighted average voltage fluctuation rate is:

[0085] ,

[0086] in, For the forgetting factor, For the The exponentially weighted average voltage fluctuation rate in time steps, Expressed as voltage, Expressed as time difference, the extracted single time step deviation and multi-time scale fluctuation rate are sorted to form a feature vector or data set for subsequent use in fuzzy controller input calculation. After that, the density clustering algorithm is introduced to dynamically determine the number of fuzzy levels N based on the preprocessed data. The formula is:

[0087]

[0088] in, is the preprocessed data, The neighborhood radius determines how far away a data point is from the neighboring data point. is the minimum number of core point samples, that is, the minimum number of data points required around a data point to be considered a core point. By adjusting these two parameters, the clustering algorithm can divide data into different numbers of fuzzy levels based on data distribution. For example, when the system is in a stable state and the data distribution is relatively concentrated, the clustering algorithm may divide the data into 3-5 levels, reducing the subsequent computational effort. However, when the system is disturbed and the data becomes dispersed, the number of levels automatically increases to 7-9, improving control accuracy.

[0089] After determining the fuzzy level, update the parameters of the membership function according to the clustering results. For the i-th fuzzy level, calculate its mean and standard deviation , then follow the formula Update the parameters of the membership function 、 、 , which can make the membership function better fit the data distribution of each fuzzy level.

[0090] In order to make full use of the advantages of different types of membership functions, we combine the triangular membership function, Gaussian membership function and trapezoidal membership function to define different types of membership functions, among which the triangular membership function:

[0091] ,

[0092] in, Parameters represented by different membership functions define the shape and position of the triangle. When \(x < a\), the membership degree of \(x\) belonging to this fuzzy set is 0, that is, it completely does not belong. When When, the membership degree linearly increases from 0 to 1 as \(x\) increases, and the increasing rate is determined by When When, the membership degree linearly decreases from 1 to 0 as \(x\) increases. When When, the membership degree becomes 0 again, that is, it completely does not belong to this fuzzy set.

[0093] Gaussian membership function:

[0094] ,

[0095] Among them, is the mean value of the Gaussian function, which determines the central position of the function, is the standard deviation of the Gaussian function. Its characteristic is that it has a smooth transition characteristic and can handle the continuous change of data. Trapezoidal membership function:

[0096] ,

[0097] Among them, \(d\) is the membership function parameter. When \(x < a\), the membership degree of \(x\) belonging to this fuzzy set is 0, that is, it completely does not belong. When When, the membership degree linearly increases from 0 to 1. When When, the membership degree remains 1. When When, the membership degree linearly decreases from 1 to 0 as \(x\) increases. When When, the membership degree becomes 0 again. The three membership functions are fused by the way of linear combination to calculate the mixed membership degree , and the formula is:

[0098] ,

[0099] Among them, are the weights of the three membership functions respectively. The particle swarm optimization algorithm is used to dynamically adjust the weights. The goal is to make the mixed membership degree as close as possible to the ideal membership degree , and the objective function is . Through continuous iterative optimization, the optimal weight combination is found. Based on the above, the calculation of the input of the fuzzy control model is completed, and the adaptive multi-modal fuzzification process is completed.

[0100] S4. Use the extracted features as inputs to construct a fuzzy control model, including constructing a fuzzy rule base, a fuzzy inference mechanism and a defuzzification module;

[0101] Using the extracted features as input, the historical operating data is analyzed using a data mining algorithm. The input data set is completely traversed, the frequency of occurrence of each item is counted, and the items are sorted in descending order of frequency. Then, the data set is scanned again, and the items in each record are inserted into the FP-tree in sorted order. Starting from the bottom of the FP-tree, frequent item sets are recursively mined. For each item, all its prefix paths in the FP-tree are found, and these prefix paths are merged into a conditional FP-tree. Frequent item sets are then mined in the conditional FP-tree. Association rules are generated based on the frequent item sets, and association rules with a confidence level greater than a threshold are selected to construct a fuzzy rule base in the form of "IF-THEN". The rules describe the relationship between different fuzzy combinations of input variables and the output control quantity. For example,

[0102] “ ", where PB, PS and NS are fuzzy language variables, representing different fuzzy states. Mamdani reasoning method is used to search for matching rules in the rule base based on the fuzzified input, and the matching degree of each matching rule is calculated:

[0103] ,

[0104] in, In order to revise the degree of membership obtained through fuzzification, evidence theory is introduced to correct the matching degree. Evidence theory can handle uncertainty and incomplete information by fusing evidence from different sources (such as historical data, real-time monitoring data, etc.) to adjust the matching degree. For example, when there are large errors in the measurement of certain input variables, evidence theory can correct the matching degree based on other relevant evidence to avoid deviations in the reasoning results due to the inaccuracy of a single piece of evidence. Finally, the weighted average defuzzification method is used to convert the fuzzy output obtained by fuzzy reasoning into an accurate control quantity. The basic idea of ​​this method is to calculate the accurate control quantity based on the rule matching degree obtained by fuzzy reasoning and the central value of the output membership function: , where n is the number of matching rules, is the rule matching degree obtained based on fuzzy reasoning, It is the central value of the output membership function, thus completing the model construction.

[0105] S5. Introducing the priority matrix and energy efficiency optimization strategy, the control quantity obtained by defuzzification is converted into the control signal of the actuator, and the actuator is intelligently coordinated controlled;

[0106] A decision tree is used to classify the importance of the relationships between different control tasks and actuators. This classification allows the construction of a priority matrix. The priority matrix is ​​a two-dimensional matrix whose rows represent different control tasks, such as temperature control, humidity control, and voltage control, and whose columns represent individual actuators, such as fans, air conditioners, humidifiers, and voltage regulators. Each element in the matrix represents the numerical priority of the corresponding actuator for a specific control task. For the temperature control task, for example, the decision tree algorithm determines the priority based on historical data on temperature deviations, fluctuation rates, and the response performance of actuators such as fans and air conditioners. A large amount of historical data is collected, including temperature setpoints, actual temperatures, temperature fluctuations, and the operating status of each actuator and its effect on temperature control. This data is used as input for the decision tree, which learns and analyzes the data to identify different situations and rules, thereby determining the priority of each actuator in temperature control. For example, if a fan can quickly and effectively adjust the temperature under certain combinations of temperature deviation and fluctuation rate, then the fan will be given a higher priority for the temperature control task in this situation; conversely, if the air conditioner is more effective under certain conditions, then the air conditioner will be given a higher priority. By performing similar analysis and classification on all control tasks and actuators, a complete priority matrix is ​​ultimately formed. This matrix serves as an important basis for the coordinated control of subsequent actuators, ensuring that the system prioritizes the appropriate actuator for different control tasks.

[0107] For each actuator, energy consumption data is collected under different operating parameters, and regression analysis is used to establish a nonlinear relationship. For example, for a fan, energy consumption data is collected under different operating parameters such as speed and air volume. Then, regression analysis methods (such as polynomial regression and neural network regression) are used to establish a nonlinear model between energy consumption and operating parameters. This model can accurately describe the energy consumption characteristics of the actuator and provide a foundation for subsequent energy efficiency optimization.

[0108] The NSGA-II (Non-Dominated Sorting Genetic Algorithm-II) algorithm is used to optimize the operating parameters of the actuators. This algorithm aims to reduce energy consumption and meet control requirements, while also considering a priority matrix. During the optimization process, the algorithm generates a set of possible operating parameter combinations, which are then evaluated and ranked according to an objective function. This objective function comprehensively considers energy consumption, control effectiveness, and the priority of the actuators. For example, for temperature control tasks, it is crucial to ensure that the temperature reaches the setpoint quickly while minimizing energy depletion, while also prioritizing actuators with high priorities. Through continuous iteration and evolution, the NSGA-II algorithm finds the optimal set of operating parameter combinations and constructs an energy efficiency optimization strategy.

[0109] The defuzzified control variables are converted into control signals for the actuators, using a distributed control system architecture. In this architecture, control tasks are distributed among multiple controllers. Each controller is responsible for controlling a portion of the actuators, and network communication enables information sharing and collaborative operation among the controllers.

[0110] For example, temperature control, humidity control, and voltage control tasks are assigned to different controllers. Each controller collaboratively controls the corresponding actuators based on a priority matrix and energy efficiency optimization strategy. When a controller receives the defuzzified control quantity for a specific control task, it first refers to the priority matrix to determine the priority order of the actuators for that task. Then, based on the optimal operating parameter combination determined by the energy efficiency optimization strategy, it converts the control quantity into a control signal that can be recognized by the specific actuator, such as a voltage signal, current signal, or pulse signal, and sends it to the corresponding actuator.

[0111] At the same time, the controllers share information through the network, providing real-time visibility into the execution status of other control tasks and the status of their actuators. If a control task experiences an anomaly or requires adjustment to the control strategy, the controllers can coordinate adjustments based on this shared information, ensuring stable operation and efficient control of the entire control cabinet system. For example, if the temperature control task requires more resources for some reason, the controllers can appropriately adjust the operating status of the actuators in the humidity control or voltage control tasks based on the priority matrix and coordination mechanism to prioritize temperature control.

[0112] S6. Optimize the model and predict faults based on the collaborative control results, and accurately adjust the control cabinet based on the prediction results;

[0113] A fitness function is constructed based on the control error and the control amount change rate. The fitness function formula is:

[0114] ,

[0115] in, It is expressed as the control error, which reflects the deviation between the actual output of the system and the expected output. represents the rate of change of the control variable, and m represents the total number of time steps. In fuzzy control, the control error is minimized because a smaller error means the system can more accurately achieve the set target. Squaring the error emphasizes the impact of larger errors and prevents positive and negative errors from canceling each other out. Sharp changes in the control variable can lead to system instability or unnecessary energy consumption. Therefore, we incorporate the rate of change of the control variable into the fitness function and multiply it by a weighting factor of 0.1 to balance the effects of the control error and the rate of change of the control variable. The association rules in the fuzzy rule base are encoded using binary coding. Based on the value of the fitness function, individuals with higher fitness are selected from the current population as parents to generate the next generation of individuals. A crossover operation is performed on these selected parents to generate new offspring individuals. The crossover operation simulates the process of gene exchange in biological genetics, exchanging chromosome segments from parent individuals to produce offspring individuals with new characteristics.

[0116] In order to make the fuzzy controller better adapt to the changes in environmental conditions, we introduce a real-time environmental compensation factor :

[0117] ,

[0118] in, Expressed as ambient temperature, Expressed as ambient humidity, Expressed as the target temperature, Expressed as target humidity, when environmental conditions change, The value of will change accordingly, providing a basis for subsequent rule weight adjustments.

[0119] The weight of the fuzzy rule is dynamically adjusted according to the environmental compensation factor. The formula is:

[0120] ,

[0121] in, is the original weight of rule i, is the adjusted weight, It is an adjustment coefficient used to control the influence of the environmental compensation factor on the rule weight. When the environmental conditions change, by adjusting the rule weight, the fuzzy controller can pay more attention to the rules related to the current environmental conditions, thereby improving the control effect. When the ambient temperature and humidity deviate from the set value by a large margin, the corresponding rule weight will be adjusted, so that the controller can control more accurately, thereby completing the precise control of the control cabinet.

[0122] Example 2

[0123] The difference between this embodiment and embodiment 1 is that this embodiment provides a precise control system for an intelligent control cabinet based on a fusion fuzzy algorithm, including:

[0124] The data acquisition module is configured to: acquire control data of the control cabinet, including environmental parameters and equipment operating status data in the control cabinet;

[0125] The preprocessing module is configured to: perform preprocessing based on the acquired control data, including using Kalman filtering for noise reduction and introducing an adaptive noise covariance adjustment mechanism;

[0126] The feature extraction module is configured to: perform feature extraction based on the preprocessed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates;

[0127] The model module is configured to: construct a fuzzy control model using the extracted features as input, including constructing a fuzzy rule base, a fuzzy inference mechanism, and a defuzzification module;

[0128] The conversion module is configured to: introduce a priority matrix and an energy efficiency optimization strategy to convert the defuzzified control quantity into a control signal for the actuator, and the actuator performs intelligent collaborative control;

[0129] The output module is configured to perform model optimization and fault prediction based on the collaborative control results, and to precisely control the control cabinet based on the prediction results.

[0130] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded and executed by a processor of a terminal device for a precise control method of an intelligent control cabinet based on a fusion fuzzy algorithm.

[0131] A terminal device includes a processor and a computer-readable storage medium, the processor is used to implement various instructions; the computer-readable storage medium is used to store multiple instructions, and the instructions are suitable for being loaded and executed by the processor to implement the method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm.

[0132] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A precise control method for an intelligent control cabinet based on a fusion fuzzy algorithm, characterized in that: include: Obtain control data of the control cabinet, including environmental parameters and equipment operating status data inside the control cabinet; Preprocessing is performed based on the acquired control data, including noise reduction using Kalman filtering and the introduction of an adaptive noise covariance adjustment mechanism; Feature extraction is performed based on the pre-processed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates; The extracted features are used as input to construct a fuzzy control model, including building a fuzzy rule base, a fuzzy reasoning mechanism and a defuzzification module; Priority matrix and energy efficiency optimization strategy are introduced to convert the control quantity obtained by defuzzification into the control signal of the actuator, and the actuator is intelligently coordinated controlled; Perform model optimization and fault prediction based on collaborative control results, and precisely control the control cabinet based on the prediction results; The feature extraction is performed based on the pre-processed control data, including calculating the temperature deviation and humidity deviation based on the pre-processed control data to extract the single time step deviation, then setting a sliding window of length L, which contains L consecutive time step data, and calculating the average voltage fluctuation rate within the window. Finally, an exponentially weighted average method is introduced to calculate the voltage fluctuation rate. The calculation formula of the exponentially weighted average voltage fluctuation rate is: ， in, For the forgetting factor, For the The exponentially weighted average voltage fluctuation rate in time steps, Expressed as voltage, Expressed as time difference.

2. The intelligent control cabinet precise control method based on fusion fuzzy algorithm according to claim 1 is characterized in that: The introduction of the adaptive noise covariance adjustment mechanism includes initializing the Kalman filter parameters according to the acquired control data, calculating the residual sequence of the data, and dynamically adjusting the process noise covariance and measurement noise covariance using the variance of the residual sequence. The adjusted covariance is combined with the new measurement information to correct the state estimation. The dynamic adjustment formula is: ， ， in, and are the adjustment coefficients for process noise covariance and measurement noise covariance, is the variance of the residual sequence, and are the process noise covariance and measurement noise covariance at the previous moment, respectively.

3. The intelligent control cabinet precise control method based on fusion fuzzy algorithm according to claim 1 is characterized in that: The feature extraction based on the pre-processed control data also includes setting membership function parameters, fuzzifying the calculated deviation and fluctuation rate, introducing a density clustering algorithm to dynamically determine the number of fuzzy levels, calculating the triangular membership function, Gaussian membership function and trapezoidal membership function according to the fuzzy levels, mixing different membership functions by dynamically adjusting weights, and converting numerical values ​​into fuzzy linguistic variables.

4. The method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm according to claim 1 is characterized in that: The method comprises the steps of: using the extracted features as input, completely traversing the input data set, counting the occurrence frequency of each item, and sorting the items in descending order of frequency; mining association rules using an FP-growth algorithm; recursively mining frequent item sets starting from the bottom of an FP-tree; generating association rules based on the frequent item sets; screening out association rules with a confidence level greater than a threshold; and constructing a fuzzy rule base in an IF-THEN format based on the screened association rules.

5. The method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm according to claim 4 is characterized in that: The construction of the fuzzy rule base, the fuzzy reasoning mechanism and the defuzzification module also includes using the Mamdani reasoning method to search for matching rules in the rule base based on the fuzzified input, calculating the matching degree of each matching rule, then introducing the evidence theory to correct the matching degree, and using the weighted average defuzzification method to calculate the control amount based on the rule matching degree obtained by fuzzy reasoning and the central value of the output membership function.

6. The method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm according to claim 1 is characterized in that: The method of performing model optimization and fault prediction based on the collaborative control results includes constructing a fitness function based on the control error and the control quantity change rate, encoding the association rules in the fuzzy rule base through binary coding, selecting individuals with higher fitness from the current population as parents to perform genetic algorithm operations based on the value of the fitness function, and then introducing a real-time environmental compensation factor, dynamically adjusting the weights of the fuzzy rules based on the environmental compensation factor to complete the model optimization.

7. An intelligent control cabinet precision control system based on fusion fuzzy algorithm, executing the method of claim 1, characterized in that: include: The data acquisition module is configured to: acquire control data of the control cabinet, including environmental parameters and equipment operating status data in the control cabinet; The preprocessing module is configured to: perform preprocessing based on the acquired control data, including using Kalman filtering for noise reduction and introducing an adaptive noise covariance adjustment mechanism; The feature extraction module is configured to: perform feature extraction based on the preprocessed control data, including extracting single time step deviations and introducing multi-time scale analysis to extract fluctuation rates; The model module is configured to: construct a fuzzy control model using the extracted features as input, including constructing a fuzzy rule base, a fuzzy inference mechanism, and a defuzzification module; The conversion module is configured to: introduce a priority matrix and an energy efficiency optimization strategy to convert the defuzzified control quantity into a control signal for the actuator, and the actuator performs intelligent collaborative control; The output module is configured to perform model optimization and fault prediction based on the collaborative control results, and to precisely control the control cabinet based on the prediction results.

8. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instructions are suitable for being loaded and executed by a processor of a terminal device according to a method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm as described in claim 1.

9. A terminal device comprising a processor and a computer-readable storage medium, wherein the processor is configured to implement various instructions; and the computer-readable storage medium is configured to store a plurality of instructions, wherein: The instructions are suitable for being loaded and executed by a processor according to a method for precise control of an intelligent control cabinet based on a fusion fuzzy algorithm as described in claim 1.

Images