Motor system energy-saving optimization automatic control system based on industrial Internet of Things

Abstract

The present invention discloses an automatic control system for energy-saving optimization of motor systems based on the Industrial Internet of Things (IIoT). This system, which relates to the field of automatic control technology, includes a data acquisition module for acquiring aeration system operating parameters, a load decoupling module for calculating the biochemical load disturbance index based on these parameters, an oxygen demand prediction module for predicting the optimal dissolved oxygen setpoint by combining the operating parameters with the disturbance index, a decision optimization module for generating optimal motor speed control instructions based on the disturbance index and the setpoint, and a model correction module for calculating the deviation factor based on the setpoint and real-time concentration, and for online correction of the oxygen demand prediction module when the threshold is exceeded. Through the collaborative efforts of multiple modules, the present invention achieves precise, energy-saving, and adaptive control of the aeration process. It can accurately distinguish loads, balance treatment effects and energy consumption, ensure long-term stable operation, and significantly reduce system energy consumption.

Description

Technical Field

[0001] The present invention relates to the field of automatic control technology, and in particular to an automatic control system for energy-saving optimization of a motor system based on the industrial Internet of Things. Background Art

[0002] Aeration system motor control often faces the challenge of processing heterogeneous data from multiple sources. Traditional control methods have limited capabilities for timestamp alignment and normalization of operational parameters such as influent water quality and flow rate, making it difficult to accurately distinguish between hydraulic and biochemical loads. Furthermore, their prediction models often use fixed parameters and lack a dynamic response mechanism to biochemical load disturbances. This makes it impossible to adaptively adjust the dissolved oxygen setpoint and motor speed based on real-time operating conditions, resulting in an imbalance between treatment effectiveness and energy consumption control when the system experiences load fluctuations.

[0003] Existing technologies suffer from significant lag in aeration system control, making it impossible to track changes in biochemical load in real time, which can easily lead to deviations in dissolved oxygen concentration control. Furthermore, traditional multi-objective optimization strategies lack a dynamic weight adjustment mechanism, making it difficult to achieve an intelligent balance between treatment effectiveness and operating energy consumption. This results in high system energy consumption and unstable treatment results. Furthermore, the lack of a model correction mechanism prevents online optimization of the prediction model based on real-time measurement data. This leads to a decrease in prediction accuracy after long-term operation, impacting system control performance. Summary of the Invention

[0004] The purpose of the present invention is to provide an automatic control system and method for energy-saving optimization of a motor system based on the industrial Internet of Things, which solves the problems existing in the background technology.

[0005] In order to solve the above technical problems, the present invention provides an automatic control system for energy-saving optimization of a motor system based on the industrial Internet of Things, comprising: a data acquisition module for obtaining operating parameters of an aeration system;

[0006] a load decoupling module, configured to calculate a biochemical load disturbance index based on the operating parameters;

[0007] an oxygen demand prediction module, configured to combine the operating parameters with the biochemical load disturbance index to predict and generate an optimal dissolved oxygen set value;

[0008] a decision optimization module, configured to generate an optimal motor speed control instruction based on the biochemical load disturbance index and the optimal dissolved oxygen set value;

[0009] The model correction module is used to calculate a model deviation factor based on the optimal dissolved oxygen set value and the collected real-time dissolved oxygen concentration, and to correct the oxygen demand prediction module online when the model deviation factor exceeds a preset deviation threshold.

[0010] Preferably, the data acquisition module is specifically used for:

[0011] Performing timestamp alignment and normalization processing on the inlet water quality parameters, inlet water flow, motor operating parameters and real-time dissolved oxygen concentration in the operating parameters;

[0012] The aligned and normalized parameters are synthesized into a unified state data vector.

[0013] Preferably, the load decoupling module is specifically used to:

[0014] Obtaining the moving average of pollutant concentrations used to characterize recent steady-state conditions;

[0015] Determining the deviation of the current influent water quality parameter from the sliding mean of the pollutant concentration;

[0016] The deviations are weighted and summed in combination with a preset weight coefficient to generate the biochemical load disturbance index.

[0017] Preferably, the oxygen demand prediction module includes:

[0018] A time series prediction model is used to infer and output the optimal dissolved oxygen set value based on the historical unified state data vector and the current biochemical load disturbance index.

[0019] Preferably, the decision optimization module is specifically used to:

[0020] Constructing a treatment effect objective function for minimizing the deviation between the real-time dissolved oxygen concentration and the optimal dissolved oxygen set value;

[0021] Constructing an operating energy consumption objective function for minimizing system energy consumption represented by the motor operating parameters;

[0022] obtaining a dynamic weight generated by a strategy function in response to the biochemical load disturbance index;

[0023] Based on the dynamic weight, performing a weighted combination of the processing effect objective function and the operation energy consumption objective function to generate a comprehensive optimization function;

[0024] The motor speed that minimizes the comprehensive optimization function under preset process constraints is solved to generate the optimal motor speed control instruction.

[0025] Preferably, the strategy function is specifically used to:

[0026] When the biochemical load disturbance index is higher than a preset disturbance threshold, increasing the weight of the treatment effect objective function;

[0027] When the biochemical load disturbance index is not higher than the preset disturbance threshold, the weight of the operation energy consumption objective function is increased.

[0028] Preferably, the step of calculating the model deviation factor by the model correction module includes:

[0029] Obtaining the model-predicted dissolved oxygen concentration generated by the oxygen demand prediction module at the previous moment;

[0030] Calculating the absolute difference between the model-predicted dissolved oxygen concentration and the currently collected real-time dissolved oxygen concentration;

[0031] The absolute difference is divided by the optimal dissolved oxygen set point to generate the model bias factor.

[0032] Preferably, the condition for triggering the online correction is:

[0033] The model deviation factor exceeds the preset deviation threshold value continuously within a preset number of cycles.

[0034] Preferably, it also includes an instruction execution module, which is used to:

[0035] Extracting a first control value from the optimal motor speed control instruction and setting it as the control instruction at the current moment;

[0036] The control instruction at the current moment is sent to the frequency converter of the motor for execution via the industrial bus.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] First, through the collaborative work of multiple modules, precise, energy-saving and adaptive control of the aeration process is achieved, which can accurately distinguish between hydraulic load and biochemical load, overcoming the lag of traditional control methods.

[0039] Secondly, each module has an innovative mechanism. The data acquisition and load decoupling module realizes accurate perception of the operating status. The oxygen demand prediction module combines historical data with the current load to generate the optimal dissolved oxygen set value. The decision optimization module balances the processing effect and energy consumption through a dynamic weight strategy. The model correction module ensures the accuracy of predictive control. The instruction execution module ensures the reliable execution of control instructions.

[0040] Third, it can reduce energy consumption, improve control accuracy and response speed while ensuring processing effects, and has functions such as fault handling and safety protection, and can operate stably for a long time. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0042] Figure 1 This is a logic block diagram of the energy-saving optimization automatic control system for the motor system based on the industrial Internet of Things of the present invention. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0044] Example 1:

[0045] See also Figure 1 , the present invention provides an automatic control system for energy-saving optimization of a motor system based on the industrial Internet of Things, comprising: a data acquisition module for obtaining operating parameters of an aeration system;

[0046] a load decoupling module, configured to calculate a biochemical load disturbance index based on the operating parameters;

[0047] an oxygen demand prediction module, configured to combine the operating parameters with the biochemical load disturbance index to predict and generate an optimal dissolved oxygen set value;

[0048] a decision optimization module, configured to generate an optimal motor speed control instruction based on the biochemical load disturbance index and the optimal dissolved oxygen set value;

[0049] The model correction module is used to calculate a model deviation factor based on the optimal dissolved oxygen set value and the collected real-time dissolved oxygen concentration, and to correct the oxygen demand prediction module online when the model deviation factor exceeds a preset deviation threshold.

[0050] The data acquisition module uses an industrial IoT sensor network to acquire real-time aeration system operating parameters, including inlet chemical oxygen demand (COD), ammonia nitrogen concentration, inlet flow rate, real-time dissolved oxygen concentration, and motor speed and power. The module aligns timestamps and normalizes these heterogeneous operating parameters to form a unified state data vector.

[0051] The load decoupling module calculates the biochemical load disturbance index by accurately distinguishing between hydraulic load and biochemical load based on the operating parameters provided by the data acquisition module. The load decoupling module obtains the sliding average of pollutant concentrations over the past T hours, calculates the deviation between the current influent water quality parameters and the sliding average, and performs a weighted summation using an adaptive weight coefficient. The biochemical load disturbance index is calculated as follows:

[0052] ;

[0053] in, : Biochemical load disturbance index, which represents the degree of influence of influent water quality fluctuation on the system;

[0054] : weight coefficient of chemical oxygen demand;

[0055] : weight coefficient of ammonia nitrogen;

[0056] : Measured value of chemical oxygen demand of forward water;

[0057] : the sliding average of the chemical oxygen demand concentration in the past T hours;

[0058] : Current measured value of influent ammonia nitrogen concentration;

[0059] : the sliding average of ammonia nitrogen concentration in the past T hours;

[0060] The weight coefficient is determined by an adaptive adjustment algorithm:

[0061] ;

[0062] in, : Historical fluctuation variance of chemical oxygen demand;

[0063] : Historical fluctuation variance of ammonia nitrogen concentration.

[0064] The load decoupling module uses fuzzy logic algorithm to identify the load type and divides the system load status into four levels: stable load, light disturbance, moderate disturbance and heavy disturbance, providing a basis for subsequent control strategy selection.

[0065] The oxygen demand prediction module combines the unified state data vector with the biochemical load disturbance index to generate the optimal dissolved oxygen setpoint using a time series prediction model based on an attention mechanism. The prediction model uses a long short-term memory network architecture and innovatively introduces an attention mechanism layer to automatically learn the importance weights of data at different times. The mathematical expression of the attention mechanism is:

[0066] ;

[0067] in, : The attention weight at time t, representing the importance of the data at that moment;

[0068] : Attention score, calculated by linear transformation and activation function;

[0069] 、 、 : parameter matrix of the attention mechanism;

[0070] : The hidden state of the model at time t;

[0071] : Model state at the last moment;

[0072] : time series length.

[0073] The output expression of the prediction model is:

[0074] ;

[0075] in, : Optimal dissolved oxygen setting value;

[0076] 、 : The weight matrix of the output layer and the hidden layer;

[0077] 、 : Bias vectors of hidden and output layers;

[0078] : context vector of attention mechanism;

[0079] : current biochemical load disturbance index;

[0080] : Linear rectification activation function.

[0081] The decision optimization module generates the optimal motor speed control command based on the biochemical load disturbance index and the optimal dissolved oxygen set value through a multi-objective optimization algorithm. The decision optimization module constructs the treatment effect objective function and the operating energy consumption objective function, and performs a weighted combination through an innovative dynamic weighting strategy:

[0082] ;

[0083] in, : Comprehensive optimization objective function;

[0084] : Final dynamic weight;

[0085] : Processing effect objective function;

[0086] : Running energy consumption objective function;

[0087] The dynamic weight strategy function adopts the form of piecewise Sigmoid function, combined with the time adaptive mechanism:

[0088] ;

[0089] ;

[0090] ;

[0091] in, : A weight factor based on load disturbance, which is adaptively adjusted with the size of BLFI;

[0092] 、 : The minimum and maximum values ​​of weight;

[0093] : slope parameter, which controls the sensitivity of weight changes;

[0094] : Load disturbance threshold, used to determine the weight switching condition;

[0095] : A time-based weight adjustment factor that reflects the impact of the diurnal cycle on energy consumption strategies;

[0096] : The current moment.

[0097] This strategy function can adaptively adjust the weight distribution of treatment effect and operating energy consumption according to the size of the biochemical load disturbance index and the characteristics of the time period, thereby achieving intelligent multi-objective balance.

[0098] The model correction module calculates the model deviation factor based on the optimal dissolved oxygen set value and real-time dissolved oxygen concentration generated by the oxygen demand prediction module:

[0099] ;

[0100] in, : Model bias factor, which measures the relative error between the predicted value and the actual value;

[0101] : The predicted dissolved oxygen value output by the oxygen demand prediction module;

[0102] : Real-time measurement of dissolved oxygen concentration;

[0103] : Optimal dissolved oxygen setting value.

[0104] The model correction module sets innovative online correction trigger conditions. When the model deviation factor exceeds the preset deviation threshold for a preset number of cycles, the online correction program is automatically triggered. The trigger conditions are:

[0105] ;

[0106] in, : Model deviation factor of the i-th period;

[0107] : The number of cycles of continuous monitoring;

[0108] : Preset the deviation threshold to determine whether the error exceeds the limit;

[0109] : Upper statistical control limit.

[0110] Online correction adopts an incremental learning algorithm, uses the experience replay mechanism to maintain an experience buffer containing recent data, and quickly adapts to new data distribution through an adaptive learning rate adjustment strategy.

[0111] The instruction execution module extracts the first control value from the control instruction sequence generated by the decision optimization module, and sends it to the motor inverter for execution through the industrial bus to implement the rolling optimization control strategy.

[0112] The Industrial Internet of Things-based motor system energy-saving optimization automatic control system achieves precise, energy-efficient, and adaptive control of the aeration process through the coordinated operation of five modules. The system accurately distinguishes between hydraulic and biochemical loads, overcoming the lag inherent in traditional control methods. This significantly reduces energy consumption while maintaining effective treatment results, achieving energy savings of over 25%.

[0113] Example 2:

[0114] The data acquisition module is specifically used for:

[0115] Performing timestamp alignment and normalization processing on the inlet water quality parameters, inlet water flow, motor operating parameters and real-time dissolved oxygen concentration in the operating parameters;

[0116] The aligned and normalized parameters are synthesized into a unified state data vector.

[0117] The load decoupling module is specifically used for:

[0118] Obtaining the moving average of pollutant concentrations used to characterize recent steady-state conditions;

[0119] Determining the deviation of the current influent water quality parameter from the sliding mean of the pollutant concentration;

[0120] The deviations are weighted and summed using preset weight coefficients to generate the biochemical load disturbance index. The data acquisition module performs timestamp alignment and normalization on the operating parameters, employing an innovative timestamp alignment algorithm to eliminate time deviations in heterogeneous data from multiple sources. This algorithm, which includes four steps: data caching, time synchronization, linear interpolation, and quality checking, identifies anomalous data by setting data change rate thresholds and physical plausibility boundaries. The data acquisition module synthesizes the aligned and normalized parameters into a unified state data vector, providing high-quality input for subsequent analysis.

[0121] The load decoupling module obtains the sliding average of pollutant concentrations and uses an exponentially weighted moving average algorithm to improve the response sensitivity to water quality changes:

[0122] ;

[0123] ;

[0124] in, 、 : exponentially weighted moving average of chemical oxygen demand and ammonia nitrogen concentration at time t;

[0125] 、 : The measured values ​​of chemical oxygen demand and ammonia nitrogen concentration at time t;

[0126] : Weighting coefficient (0< ≤1), controls the response sensitivity to the current data.

[0127] The load decoupling module determines the deviation between the current influent water quality parameters and the sliding average, and generates a biochemical load disturbance index based on the adaptive weight coefficient. The weight coefficient is adjusted according to the water quality characteristics of different seasons. In spring and summer, the weight of chemical oxygen demand is 0.7, the weight of ammonia nitrogen is 0.3, and in autumn and winter, the weights of both are 0.5.

[0128] The load decoupling module uses a fuzzy logic algorithm to identify load types. Based on the magnitude and trend of the biochemical load disturbance index, a trapezoidal membership function is used to classify load states into four levels. This identification result guides subsequent control strategy selection and parameter adjustment, achieving a load identification accuracy exceeding 95%.

[0129] The collaborative work of the data acquisition module and the load decoupling module realizes the precise perception of the operating status of the aeration system and the accurate identification of the load characteristics. The data synchronization accuracy reaches the second level, providing a key decision-making basis for the adaptive control of the system.

[0130] Example 3:

[0131] The oxygen demand prediction module includes:

[0132] A time series prediction model is used to infer and output the optimal dissolved oxygen set value based on the historical unified state data vector and the current biochemical load disturbance index.

[0133] The oxygen demand forecasting module includes a time series prediction model based on an attention mechanism. This model innovatively combines historical unified state data vectors with the current biochemical load disturbance index to infer and output the optimal dissolved oxygen setpoint. The time series prediction model utilizes a long-short-term memory network architecture and incorporates an attention mechanism layer to automatically learn the importance of data at different moments, enhancing the model's focus on information at critical moments.

[0134] The attention mechanism layer adopts the self-attention mechanism, and the mathematical expression is:

[0135] ;

[0136] in, : the attention weight of the data at time t in the time series;

[0137] : The intermediate value that measures the importance of the data at time t, generated by linear transformation and activation function;

[0138] : parameter vector of the attention mechanism, used to map the transformed hidden state to a scalar score;

[0139] 、 : Weight matrix, respectively for the current moment hidden state and the previous state Perform linear transformation;

[0140] : The hidden state of the long short-term memory network (LSTM) at time t, which contains the feature information at that moment;

[0141] : The model state at time t−1, used to capture the dependencies of the time series.

[0142] The innovative output expression of the time series forecasting model is:

[0143] ;

[0144] in, : context vector;

[0145] : biochemical load disturbance index;

[0146] 、 : weight matrix;

[0147] 、 : bias vector.

[0148] The context vector Biochemical load disturbance index The splicing realizes the organic combination of historical information and current load status, which is the key innovation of the present invention.

[0149] The time series prediction model is trained offline using extensive historical operational data, employing a mean squared error loss function and regularization techniques to prevent overfitting. In actual operation, the model predicts the optimal dissolved oxygen setpoint within the next 15-30 minutes based on the current state in real time, achieving over 97% accuracy.

[0150] The oxygen demand prediction module also includes a model performance monitoring submodule, which evaluates the performance indicators of the prediction model such as root mean square error, mean absolute error and determination coefficient in real time, providing a basis for model correction.

[0151] When faced with a sudden change in the inlet chemical oxygen demand concentration, the oxygen demand prediction module can predict the adjustment requirements for the optimal dissolved oxygen set point within 3 minutes. Compared with the traditional PID control method, the response time is shortened by more than 60%, and the energy saving effect is improved by more than 18%.

[0152] Example 4:

[0153] The decision optimization module is specifically used to:

[0154] Constructing a treatment effect objective function for minimizing the deviation between the real-time dissolved oxygen concentration and the optimal dissolved oxygen set value;

[0155] Constructing an operating energy consumption objective function for minimizing system energy consumption represented by the motor operating parameters;

[0156] obtaining a dynamic weight generated by a strategy function in response to the biochemical load disturbance index;

[0157] Based on the dynamic weight, performing a weighted combination of the processing effect objective function and the operation energy consumption objective function to generate a comprehensive optimization function;

[0158] The motor speed that minimizes the comprehensive optimization function under preset process constraints is solved to generate the optimal motor speed control instruction.

[0159] The strategy function is specifically used to:

[0160] When the biochemical load disturbance index is higher than a preset disturbance threshold, increasing the weight of the treatment effect objective function;

[0161] When the biochemical load disturbance index is not higher than the preset disturbance threshold, the weight of the operation energy consumption objective function is increased.

[0162] The decision-making optimization module constructs a treatment performance objective function and an operating energy consumption objective function, innovatively employing a dynamic weighting strategy for weighted combination. The treatment performance objective function adopts a weighted quadratic form, not only considering the tracking accuracy of dissolved oxygen concentration but also constraining the rate of change of dissolved oxygen concentration. The operating energy consumption objective function is based on the product of motor power and operating time, comprehensively considering the impact of changes in motor speed on energy consumption.

[0163] The decision optimization module obtains the dynamic weights generated by the strategy function in response to the biochemical load disturbance index. The strategy function adopts the innovative piecewise Sigmoid function form, which can adaptively adjust the weight distribution between treatment effect and operating energy consumption according to the size of the biochemical load disturbance index:

[0164] ;

[0165] in, : Weight function based on biochemical load disturbance;

[0166] : biochemical load disturbance index;

[0167] : load disturbance threshold;

[0168] : Minimum weight value, corresponding to priority energy consumption optimization under low load;

[0169] : Maximum weight value, corresponding to the priority processing effect under high load;

[0170] : slope parameter, which controls the sensitivity of the weight to changes in BLFI;

[0171] The policy function also includes a time-adaptive mechanism to adjust the weight preference according to different times of the day:

[0172] ;

[0173] in, : Time adaptive mechanism;

[0174] : current moment;

[0175] The final dynamic weight is:

[0176] ;

[0177] in, : The final dynamic weight;

[0178] This strategy function has a smooth transition characteristic, which avoids the impact of weight mutation on system stability and achieves an intelligent balance between processing effect and operating energy consumption.

[0179] The decision optimization module generates a comprehensive optimization function based on dynamic weights:

[0180] ;

[0181] in, : Comprehensive optimization objective function;

[0182] : Final dynamic weight;

[0183] : Processing effect objective function;

[0184] :Running energy consumption objective function; and Normalization is performed, for example, mapping its value to the interval [0, 1], to eliminate the dimension effect and make the weighted sum have clear physical meaning.

[0185] The decision optimization module uses a sequential quadratic programming algorithm to solve the constrained optimization problem and generates the optimal motor speed control instruction while meeting process constraints such as motor speed limit and dissolved oxygen concentration lower limit.

[0186] In actual application, when the influent quality is stable, the strategy function sets the dynamic weight to 0.3, prioritizing energy-saving operation and reducing energy consumption by over 22%. When the biochemical load disturbance index exceeds the threshold, the strategy function automatically adjusts the dynamic weight to 0.8, prioritizing treatment effectiveness and ensuring that the effluent quality remains stable and meets standards. This adaptive control strategy improves control accuracy by over 30% and response speed by over 50%.

[0187] Example 5:

[0188] The step of calculating the model deviation factor by the model correction module includes:

[0189] Obtaining the model-predicted dissolved oxygen concentration generated by the oxygen demand prediction module at the previous moment;

[0190] Calculating the absolute difference between the model-predicted dissolved oxygen concentration and the currently collected real-time dissolved oxygen concentration;

[0191] The absolute difference is divided by the optimal dissolved oxygen set point to generate the model bias factor.

[0192] The conditions for triggering the online correction are:

[0193] The model deviation factor exceeds the preset deviation threshold value continuously within a preset number of cycles.

[0194] The model correction module obtains the model-predicted dissolved oxygen concentration generated by the oxygen demand prediction module at the previous moment and uses a sliding window filtering algorithm to preprocess the real-time dissolved oxygen concentration to eliminate the influence of measurement noise. The model correction module calculates the absolute difference between the model-predicted dissolved oxygen concentration and the real-time measured value, and then divides this difference by the optimal dissolved oxygen set value to generate the model deviation factor:

[0195] ;

[0196] in, : Model bias factor, which measures the relative error between the predicted value and the actual value;

[0197] : The predicted dissolved oxygen value output by the oxygen demand prediction module;

[0198] : Real-time measurement of dissolved oxygen concentration;

[0199] : Optimal dissolved oxygen setting value;

[0200] The model correction module uses statistical process control methods to monitor the changing trend of the model deviation factor, sets upper and lower control limits based on control chart theory, and monitors the stability of model performance in real time.

[0201] The model correction module sets an innovative online correction trigger condition. When the model deviation factor exceeds the preset deviation threshold for a preset number of cycles, the online correction is triggered. The mathematical expression of the trigger condition is:

[0202] ;

[0203] This trigger condition takes into account both the absolute threshold and the statistical control limit, avoiding false triggering due to accidental measurement noise and improving the reliability of trigger judgment.

[0204] The model correction module also sets emergency correction trigger conditions. When the model deviation factor exceeds the emergency threshold, online correction is triggered immediately without waiting for continuous cycle judgment.

[0205] When online correction is triggered, the model correction module automatically starts the model parameter update program and uses the incremental learning algorithm to perform online training on the time series prediction model. The incremental learning algorithm uses the experience replay mechanism to maintain an experience buffer containing recent data, and new samples replace old samples on a first-in-first-out basis. The objective function of incremental learning is:

[0206] ;

[0207] in, : The total loss function of incremental learning;

[0208] : weight parameter;

[0209] : loss function for new data;

[0210] : loss function of old data;

[0211] The incremental learning algorithm can quickly adapt to the new data distribution without losing historical knowledge, and the learning rate adopts an adaptive adjustment strategy to dynamically adjust according to the changes in the loss function.

[0212] The model correction module also updates key parameters in the biochemical reaction kinetics model and uses recursive least squares to determine the optimal parameter values ​​by minimizing the weighted sum of squares of prediction errors.

[0213] The model correction module ensures the system maintains high-precision prediction and control performance through continuous performance monitoring and adaptive correction. Over long-term operation, the model's prediction accuracy remains above 92%, with online corrections triggered an average of 2-3 times per month. Each correction improves model accuracy by 5-8%, and corrections take less than 30 minutes.

[0214] Example 6:

[0215] Also included is an instruction execution module, which is used to:

[0216] Extracting a first control value from the optimal motor speed control instruction and setting it as the control instruction at the current moment;

[0217] The control instruction at the current moment is sent to the frequency converter of the motor for execution via the industrial bus.

[0218] The instruction execution module extracts the first control value from the optimal motor speed control instruction sequence generated by the decision optimization module as the current control instruction. The instruction execution module uses a multi-level safety check mechanism to ensure the safety of control instructions, including value range checks, rate of change limit checks, and system status checks.

[0219] The instruction execution module converts the data format of the control instruction and uses a high-precision quantization algorithm to convert the floating-point speed value into the integer digital value required by the inverter:

[0220] ;

[0221] in, : converted digital control instruction;

[0222] : The optimal motor speed control value at the current moment;

[0223] 、 : Minimum and maximum values ​​of motor speed;

[0224] : The resolution of the inverter digital quantity;

[0225] round() is a standard mathematical function that rounds a value to the nearest integer;

[0226] The conversion accuracy is controlled within ±0.1rpm, and the converted digital quantity is checked and calculated to ensure data transmission integrity.

[0227] The command execution module sends control commands to the motor inverter via the industrial bus for execution. It utilizes a redundant communication mechanism to enhance communication reliability, including primary and backup communication links and a retransmission mechanism. After sending control commands, the command execution module verifies command execution by reading the inverter status register and regularly monitors the actual operating status of the motor.

[0228] The instruction execution module uses a state feedback control algorithm to make fine adjustments based on the deviation between the actual speed and the set speed:

[0229] ;

[0230] in, : Adjusted motor speed command;

[0231] 、 : Set speed and actual speed;

[0232] 、 : Proportional and integral control coefficients, used to adjust the speed deviation;

[0233] The instruction execution module has fault handling and safety protection functions. When an inverter fault or communication abnormality is detected, it automatically switches to safety mode, sets the motor speed to a preset safety value, and sends a fault alarm message.

[0234] The instruction execution module is equipped with a data recording function, which records the execution of control instructions and system operating status in real time, providing data support for system maintenance and performance analysis.

[0235] In practical applications, the command execution module achieves precise control command issuance and reliable execution, with a control command response time of less than 500ms, motor speed control accuracy reaching ±0.5rpm, and a communication success rate exceeding 99.9%. The system has operated continuously for more than 8,760 hours, with downtime less than 0.1%, and cumulative energy savings exceeding 20%.

[0236] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. The motor system energy-saving optimization automatic control system based on industrial Internet of Things is characterized by: include: Data acquisition module, used to obtain the operating parameters of the aeration system; a load decoupling module, configured to calculate a biochemical load disturbance index based on the operating parameters; an oxygen demand prediction module, configured to combine the operating parameters with the biochemical load disturbance index to predict and generate an optimal dissolved oxygen set value; a decision optimization module, configured to generate an optimal motor speed control instruction based on the biochemical load disturbance index and the optimal dissolved oxygen set value; a model correction module, configured to calculate a model deviation factor based on the optimal dissolved oxygen set value and the collected real-time dissolved oxygen concentration, and to correct the oxygen demand prediction module online when the model deviation factor exceeds a preset deviation threshold; The load decoupling module is specifically used for: Obtaining the moving average of pollutant concentrations used to characterize recent steady-state conditions; Determining the deviation of the current influent water quality parameter from the sliding mean of the pollutant concentration; Performing weighted summation on the deviations in combination with a preset weight coefficient to generate the biochemical load disturbance index; The oxygen demand prediction module includes: A time series prediction model for inferring and outputting the optimal dissolved oxygen set value based on a historical unified state data vector and a current biochemical load disturbance index; The decision optimization module is specifically used to: Constructing a treatment effect objective function for minimizing the deviation between the real-time dissolved oxygen concentration and the optimal dissolved oxygen set value; Constructing an operating energy consumption objective function for minimizing system energy consumption represented by the motor operating parameters; obtaining a dynamic weight generated by a strategy function in response to the biochemical load disturbance index; Based on the dynamic weight, performing a weighted combination of the processing effect objective function and the operation energy consumption objective function to generate a comprehensive optimization function; Solving the motor speed that minimizes the comprehensive optimization function under preset process constraints to generate the optimal motor speed control instruction; The step of calculating the model deviation factor by the model correction module includes: Obtaining the model-predicted dissolved oxygen concentration generated by the oxygen demand prediction module at the previous moment; Calculating the absolute difference between the model-predicted dissolved oxygen concentration and the currently collected real-time dissolved oxygen concentration; The absolute difference is divided by the optimal dissolved oxygen set point to generate the model bias factor.

2. The motor system energy-saving optimization automatic control system based on industrial Internet of Things according to claim 1 is characterized in that: The data acquisition module is specifically used for: Performing timestamp alignment and normalization processing on the inlet water quality parameters, inlet water flow, motor operating parameters and real-time dissolved oxygen concentration in the operating parameters; The aligned and normalized parameters are synthesized into a unified state data vector.

3. The motor system energy-saving optimization automatic control system based on industrial Internet of Things according to claim 1 is characterized in that: The strategy function is specifically used to: When the biochemical load disturbance index is higher than a preset disturbance threshold, increasing the weight of the treatment effect objective function; When the biochemical load disturbance index is not higher than the preset disturbance threshold, the weight of the operation energy consumption objective function is increased.

4. The motor system energy-saving optimization automatic control system based on industrial Internet of Things according to claim 1 is characterized in that: The conditions for triggering the online correction are: The model deviation factor exceeds the preset deviation threshold value continuously within a preset number of cycles.

5. The motor system energy-saving optimization automatic control system based on industrial Internet of Things according to claim 1 is characterized in that: Also included is an instruction execution module, which is used to: Extracting a first control value from the optimal motor speed control instruction and setting it as the control instruction at the current moment; The control instruction at the current moment is sent to the frequency converter of the motor for execution via the industrial bus.

Images