Electric pump control method and system based on instruction perception

Abstract

The invention discloses an electric pump control method and system based on instruction perception, and relates to the technical field of electric pump control, comprising: starting a voice recognition module, a wireless communication module and a sensor acquisition module, and establishing a bidirectional communication link with a remote controller; an AI-driven noise reduction algorithm is adopted to pre-process the environment audio collected by the microphone array, and an improved hidden Markov model is utilized to perform keyword recognition on the pre-processed audio to obtain a voice instruction signal; the remote controller sends a control instruction to a receiving end, analyzes the instruction content, records the current communication channel quality, dynamically adjusts the transmitting power, the frequency hopping strategy or the coding mode according to the channel quality, and ensures stable communication between the remote controller and the electric pump; and the main control unit simultaneously receives a voice command signal and a remote controller command signal, sets a command priority rule in combination with data acquired by the sensor, and executes corresponding operation according to the priority.

Description

Technical Field

[0001] The present invention relates to the technical field of electric pump control, and particularly to an electric pump control method and system based on instruction perception. Background Art

[0002] Electric pump control technology refers to the technology of precisely regulating the working state of an electric pump through an electronic control system. Its main goal is to effectively control key parameters such as the pressure and flow rate at the pump outlet to meet the requirements of different application scenarios. Therefore, how to use advanced technical means to improve the intelligent level and safety of electric pump control has become one of the urgent problems to be solved currently.

[0003] In the field of electric pump control, existing electric pump systems simultaneously receive control signals from multiple sources such as remote controllers, voice commands, and automatic control systems. Without distinguishing the priority order, it is easy to cause command conflicts or response chaos. Moreover, traditional control systems mostly rely on real-time feedback regulation and cannot predict potential risks in advance, such as sudden pressure changes, abnormal flow rates, or excessive temperature rises. At the same time, traditional PID controllers have problems such as regulation lag and serious overshoot when facing non-linear and time-varying systems, which affect the control performance. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an electric pump control method based on instruction perception to solve the problems that traditional control systems mostly rely on real-time feedback regulation and cannot predict potential risks in advance, such as sudden pressure changes, abnormal flow rates, or excessive temperature rises. At the same time, traditional PID controllers have problems such as regulation lag and serious overshoot when facing non-linear and time-varying systems, which affect the control performance.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides an electric pump control method based on instruction perception, which includes: Start the voice recognition module, wireless communication module, and sensor acquisition module, and establish a two-way communication link with the remote controller; Preprocess the ambient audio collected through the microphone array using an AI-driven noise reduction algorithm, and perform keyword recognition on the preprocessed audio using an improved hidden Markov model to obtain a voice command signal; The remote controller sends a control command to the receiving end, parses the command content, and records the current communication channel quality. Dynamically adjust the transmission power, frequency hopping strategy, or coding method according to the channel quality to ensure stable communication between the remote controller and the electric pump; The main control unit simultaneously receives voice command signals and remote control command signals, combines the data collected by the sensors, sets the command priority rules, and performs corresponding operations according to the priority. According to the command content, a PWM signal is sent to the motor controller to adjust the motor speed, and the pressure output or flow rate of the plunger pump is controlled. The pressure, flow rate, and temperature parameters are collected in real time, and the data is uploaded to the main control unit after filtering processing. The machine learning model is used for trend prediction to determine whether there are abnormal fluctuations. Based on the sensor feedback data, the main control unit dynamically adjusts the motor speed and the pump outlet pressure, and uses the PID control algorithm combined with fuzzy logic to optimize the control curve to achieve dynamic feedback regulation. The display real-time displays the current working status and provides a live view through the camera. If an emergency occurs, a warning screen will automatically pop up and a voice reminder will be played.

[0007] As a preferred solution of the electric pump control method based on instruction perception according to the present invention, wherein: the improved hidden Markov model performs keyword recognition on the preprocessed audio to obtain the voice command signal, and the specific steps are as follows: The received environmental audio signal is subjected to time-domain enhancement processing to obtain a denoised pure audio signal. The audio signal is subjected to short-time Fourier transform STF, and the Mel frequency cepstral coefficients of each frame are extracted to form an acoustic feature vector sequence. The improved hidden Markov model is defined as a five-tuple, and the expression is: ; Among them, is the initial state distribution, is the state transition matrix, is the observation probability matrix, is the attention weight set, is the state transition weighting function, which is used to dynamically adjust the state transition probability; The attention weight is introduced to enhance the semantic connection between adjacent frames, and the expression is: ; Among them, is the cosine similarity between the frame and the previous frame, is the attention scaling factor, is the frame's attention weight, indicating the importance of this frame in the overall recognition process; The state transition weighting function is defined, and the expression is: ; Among them, is the total number of hidden states in the model, is the state jump suppression factor, represents state and state the distance between; The state transition weighting function is used to penalize cross-state jumps and encourage the model to recognize speech commands along a reasonable path; Multiply the state transition matrix in the original HMM by the weighting function to obtain a new state transition matrix; Based on the original observation probability, introduce the attention weight to weight the observation values of each frame and update the observation probability matrix; Use the Viterbi algorithm to decode the improved hidden Markov model, find the hidden state sequence, and map it to the keyword recognition result; The expression of the Viterbi algorithm is: ; where, represents the maximum path probability of being in state at time , is the updated state transition probability, is the updated observation probability, is the observation value of the The finally output recognized keyword set , which includes: start, stop, and speed regulation, and send it as a voice command signal to the main control unit.

[0008] As a preferred scheme of the electric pump control method based on instruction perception described in the present invention, wherein: combining the data collected by the sensor, setting an instruction priority rule, and performing corresponding operations according to the priority, the specific steps are: Use a pressure sensor to monitor the pressure at the pump outlet end in real time; Use a flow sensor to measure the volume of liquid flowing through the pipeline; Use a temperature sensor to collect the working temperature of the motor and the hydraulic system; Denoise the collected original data through a low-pass filter respectively to obtain a smoothed sensor signal sequence; Define an anomaly detection function to identify whether the current working condition is in a dangerous state, and the expression is:

[0009]

[0010] ; Among them, is the maximum allowable pressure value, , are the minimum and maximum allowable flow ranges respectively, is the highest allowable temperature of the motor or hydraulic system, , , are Boolean flags indicating whether pressure, flow, and temperature exceed the safety threshold respectively; Receives three types of control inputs, including: remote control instructions from the remote controller, voice instructions from the voice recognition module, and automatic control instructions generated by the main control unit based on sensor feedback; Defines an instruction priority function, and the expression is: ; Among them, is the instruction priority score, indicates whether it is from the safety system, indicates whether it is from the remote controller, indicates whether it is from the voice instruction, > > , is the weight coefficient; At each moment , the main control unit simultaneously receives , , Three instructions, and calculate their corresponding priority scores respectively , , ; Compare the magnitudes of the three, and select the instruction corresponding to the largest one as the final execution instruction; Convert the selected final execution instruction into specific control signals, including: sending a PWM waveform to the motor controller to adjust the motor speed, controlling the solenoid valve to adjust the output pressure or flow of the plunger pump, and triggering the alarm device or the display to update the working state information.

[0011] As a preferred solution of the electric pump control method based on instruction perception according to the present invention, wherein: the pressure, flow, and temperature parameters are collected in real time, and the data is uploaded to the main control unit after filtering processing, and a machine learning model is used for trend prediction to determine whether there is abnormal fluctuation. The specific steps are as follows: Use the smoothed sensor signal sequence as the input variable of the trend prediction model; Define the length of the time window centered on the current moment as, extract the historical information of the past N frames from the smoothed sensor signal sequence, and construct a multi-dimensional time series feature vector; Adopt the long short-term memory network LSTM as the trend prediction model. Its structure includes an input layer, a hidden layer, and an output layer. The input layer receives the feature vector, the hidden layer consists of multiple LSTM cells, and the output layer generates the predicted value for the next moment. The expression is: ; Among them, represents the LSTM neural network mapping function, is the set of model parameters, including the weight matrix and the bias term, is for the next time period the predicted value; In the model training stage, use the historical sensor data set and minimize the mean square error loss function. The expression is: ; Among them, is the actual observed value, is the model predicted value, is the overall prediction error of the model; Define the residual function to represent the deviation between the current predicted value and the actual value. The expression is: ; Among them, is the actual sensor value at the current moment, is the predicted value of the model for the current moment, is the prediction error; Further introduce the moving average residual and standard deviation to adaptively adjust the anomaly detection threshold; For the sensor parameters, perform the above prediction and anomaly detection processes respectively, and output two types of information, including: the trend prediction result and the anomaly status flag.

[0012] As a preferred solution of the electric pump control method based on instruction perception described in the present invention, among them: the main control unit dynamically adjusts the motor speed and the pump outlet pressure according to the sensor feedback data, and uses the PID control algorithm combined with fuzzy logic to optimize the control curve to achieve dynamic feedback regulation. The specific steps are: Use a pressure sensor to collect the actual pressure at the outlet end of the piston pump in real time; A flow sensor collects the actual flow rate flowing through the system; A temperature sensor collects the motor winding temperature and the hydraulic oil temperature; Set the corresponding target pressure, target flow rate, and safety temperature upper limit according to the upper layer control instruction or the preset working condition; Define two fuzzy input variables of pressure error and pressure error change rate, and at the same time define the fuzzy output variable of the PID parameter gain correction factor to construct a membership function set; Define the standard PID control law, and the expression is: ; Among them, is the output of the controller, , , are the basic proportional, integral, and differential coefficients respectively, is the current pressure error, is the historical error sequence; Introduce the fuzzy logic output to perform weighted adjustment on the PID parameters to obtain the improved control law, and the expression is: ; Among them, is the improved control law; Convert the output by the improved PID controller into a PWM waveform with adjustable duty cycle to control the motor frequency converter, and then adjust the motor speed; By changing the motor speed, dynamically adjust the pressure output and flow characteristics of the plunger pump.

[0013] As a preferred solution of the electric pump control method based on instruction perception according to the present invention, wherein: in the dynamic feedback regulation, a temperature protection mechanism is further introduced to achieve multi-dimensional collaborative control, including: Define the temperature anomaly flag function, and the expression is: ; When , trigger the temperature protection mechanism and force the PWM duty cycle to be lowered, as expressed below: ; Among them, is the temperature suppression coefficient, is the highest tolerable temperature of the system, is the PWM duty cycle after temperature correction.

[0014] As a preferred solution of the electric pump control method based on instruction perception according to the present invention, wherein: when an emergency occurs, an alarm screen is automatically popped up and a voice reminder is played. The specific steps are as follows: Define the event trigger function, and the expression is: ; Among them, is the event trigger flag. If any sensor anomaly flag is activated, it is determined as an emergency event; When = 1, execute actions, and the execution actions include: popping up a red warning box on the main interface of the display, starting the speech synthesis module, playing a preset voice reminder statement, writing the sensor data and abnormal information at the current moment into the local memory, and sending an alarm signal to the management terminal through the wireless communication module; The voice reminder content is dynamically selected according to the abnormal type, and the expression is as follows: ; Among them, is the final synthesized voice content, is the basic statement template, , , is the conditional splicing statement segment.

[0015] In a second aspect, the present invention provides an electric pump control system based on instruction perception, including: A voice recognition module, a communication management module, a priority decision module, a status monitoring module, a dynamic regulation module, and a human-computer interaction module; The voice recognition module is used to start the microphone array and preprocess the environmental audio using an AI-driven noise reduction algorithm, and perform keyword recognition on the audio signal using an improved hidden Markov model, and output a voice command signal; The communication management module is used to establish a two-way communication link with the remote control, parse the control instruction content, and dynamically adjust the transmission power, frequency hopping strategy, or coding method according to the communication channel quality to ensure stable communication; The priority decision module is used to receive the voice command signal and the remote control command signal, combine the pressure, flow rate, and temperature data collected by the sensor, set the multi-source instruction priority rule, and select and execute the corresponding operation according to the priority; The status monitoring module is used to collect pressure, flow rate, and temperature parameters in real time, upload them to the main control unit after filtering, and use the LSTM machine learning model for trend prediction to determine whether there is abnormal fluctuation; The dynamic regulation module is used to dynamically adjust the motor speed and the pump outlet pressure according to the sensor feedback data, adopt the PID control algorithm combined with fuzzy logic to optimize the control curve, realize closed-loop feedback regulation, and introduce a temperature protection mechanism to prevent overheating damage; The human-computer interaction module is used to display the current working status on the display in real time, provide a live view through the camera, automatically pop up a warning screen and play a voice reminder when an emergency is detected, and at the same time support remote alarm notification and log recording.

[0016] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the instruction perception-based electric pump control method described in the first aspect of the present invention is implemented.

[0017] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the instruction perception-based electric pump control method described in the first aspect of the present invention is implemented.

[0018] The beneficial effects of the present invention are as follows: By introducing an AI-driven noise reduction algorithm to perform time-domain enhancement processing on the original audio, environmental noise is effectively suppressed, and the clarity of the speech signal is improved. By constructing an improved hidden Markov model that combines an attention mechanism and a state transition weighting function, the semantic correlation between adjacent speech frames is enhanced, and the state transition path selection process is optimized, significantly improving the accuracy and robustness of keyword recognition. By constructing a priority scoring function including weight coefficients, dynamically evaluating the importance of different instructions in combination with sensor feedback data, and using a maximum value selection strategy to determine the final executed instruction, it is ensured that the system can give priority to responding to the operation with the highest safety and strongest rationality in any situation, realizing an automatic priority intervention mechanism in case of emergency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 It is a flowchart of the instruction perception-based electric pump control method in Embodiment 1.

[0021] Figure 2 It is a schematic diagram of the instruction perception-based electric pump control system in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0023] Many specific details are set forth in the following description to facilitate a thorough understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" mentioned herein refers to specific features, structures, or characteristics that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or selectively exclusive embodiments from other embodiments.

[0025] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a method for controlling an electric pump based on instruction perception, including the following steps: S1. Start the microphone array and preprocess the ambient audio using an AI-driven noise reduction algorithm. Use an improved hidden Markov model to identify keywords in the audio signal and output a voice instruction signal; Furthermore, perform time-domain enhancement processing on the received ambient audio signal to obtain a denoised pure audio signal; Perform a short-time Fourier transform (STF) on this audio signal, extract the Mel frequency cepstral coefficients of each frame, and form an acoustic feature vector sequence; Define the improved hidden Markov model as a five-tuple, and the expression is: ; Among them, is the initial state distribution, is the state transition matrix, is the observation probability matrix, is the set of attention weights, is the state transition weighting function, used to dynamically adjust the state transition probability; Introduce attention weights to enhance the semantic connection between adjacent frames. The expression is: ; Among them, is the cosine similarity between the th frame and the previous frame, is the attention scaling factor, is the th frame's attention weight, indicating the importance of this frame in the overall recognition process; Define the state transition weighting function, and the expression is: ; Among them, is the total number of hidden states in the model, is the state jump suppression factor, represents the state and the state distance between; The state transition weighting function is used to penalize cross-state jumps and encourage the model to identify speech commands along reasonable paths; Multiply the state transition matrix in the original HMM by the weighting function to obtain a new state transition matrix; Based on the original observation probability, introduce attention weights to weight the observed values of each frame and update the observation probability matrix; Use the Viterbi algorithm to decode the improved hidden Markov model, find the hidden state sequence, and map it to the keyword recognition result; The expression of the Viterbi algorithm is: ; where represents the maximum path probability of being in state at time , is the updated state transition probability, is the updated observation probability, is the th frame of the observed value; Finally, output the recognized keyword set , which includes: start, stop, and speed regulation, and send it as a voice command signal to the main control unit; It should be noted that the improved hidden Markov model combined with the attention mechanism and the state transition weighting function effectively improves the robustness and accuracy of speech recognition in complex industrial environments. This model structure is suitable for embedded deployment, with the characteristics of low latency and high precision, and can meet the dual requirements of real-time and stability of the electric pump system, and is suitable for scenarios such as remote operation and unattended operation.

[0026] S2. Receive the voice command signal and the remote control command signal, combine the pressure, flow rate, and temperature data collected by the sensor, set the multi-source command priority rule, and select and execute the corresponding operation according to the priority; Furthermore, a pressure sensor is used to monitor the pressure at the pump outlet end in real time; A flow sensor is used to measure the volume of the liquid flowing through the pipeline; A temperature sensor is used to collect the working temperature of the motor and the hydraulic system; The collected original data is respectively denoised by a low-pass filter to obtain a smoothed sensor signal sequence; Define an anomaly detection function to identify whether the current working condition is in a dangerous state, and the expression is:

[0027]

[0028] ; wherein, is the maximum allowable pressure value, , are the minimum and maximum allowable flow rate ranges respectively, is the maximum allowable temperature of the motor or hydraulic system, , , are Boolean flags indicating whether pressure, flow rate, and temperature exceed the safety thresholds respectively; Receives three types of control inputs, including: remote control instructions from a remote controller, voice instructions from a voice recognition module, and automatic control instructions generated by the main control unit based on sensor feedback; Defines an instruction priority function, and the expression is: ; wherein, is the instruction priority score, indicates whether it is from the safety system, indicates whether it is from the remote controller, indicates whether it is from the voice instruction, > > , is the weight coefficient; At each moment , the main control unit simultaneously receives , , three instructions, and calculates their corresponding priority scores respectively , , ; Compare the magnitudes of the three, and select the instruction corresponding to the largest one as the final execution instruction; Convert the selected final execution instruction into specific control signals, including: sending a PWM waveform to the motor controller to adjust the motor speed, controlling the solenoid valve to adjust the output pressure or flow rate of the plunger pump, and triggering the alarm device or the display to update the working state information; It should be noted that by setting the multi-source instruction priority rules, intelligent decision-making and orderly response between remote control instructions, voice instructions, and automatic control instructions are achieved. This mechanism ensures that automatic control instructions have the highest priority under abnormal working conditions, thereby preventing safety accidents caused by human misoperation or communication interference, and improving the fault tolerance and operation stability of the system, especially suitable for high-risk working environments.

[0029] S3. Real-time collect pressure, flow rate, and temperature parameters, upload them to the main control unit after filtering, and use the LSTM machine learning model for trend prediction to determine whether there are abnormal fluctuations; Furthermore, use the smoothed sensor signal sequence as the input variable of the trend prediction model; Define the time window length centered on the current moment as, extract the historical information of the past N frames from the smoothed sensor signal sequence, and construct a multi-dimensional time series feature vector; Adopt the long short-term memory network LSTM as the trend prediction model. Its structure includes an input layer, a hidden layer, and an output layer. The input layer receives the feature vector, the hidden layer consists of multiple LSTM units, and the output layer generates the predicted value for the next moment. The expression is: ; Among them, represents the LSTM neural network mapping function, is the set of model parameters, including the weight matrix and the bias term, is for the next time period 's predicted value; In the model training stage, use the historical sensor data set and minimize the mean square error loss function. The expression is: ; Among them, is the actual observed value, is the model predicted value, is the overall prediction error of the model; Define the residual function to represent the deviation between the current predicted value and the actual value. The expression is: ; Among them, is the actual sensor value at the current moment, is the predicted value of the model for the current moment, is the prediction error; Further introduce the moving average residual and standard deviation to adaptively adjust the anomaly detection threshold; For the sensor parameters, respectively execute the above prediction and anomaly detection processes, and output two types of information, including: trend prediction results and anomaly status flags; It should be noted that the trend prediction model constructed based on the LSTM network can accurately predict the future change trends of pressure, flow rate, and temperature parameters without relying on manual threshold setting, and combine the moving average residual and standard deviation mechanisms to achieve adaptive anomaly detection. The method significantly improves the system's early fault identification ability, provides data support for feedforward control and preventive maintenance, and helps to improve the reliability of equipment operation and the efficiency of operation and maintenance.

[0030] S4. Based on sensor feedback data, the PID control algorithm is combined with fuzzy logic to optimize the control curve, dynamically adjust the motor speed and pump outlet pressure, implement closed-loop feedback regulation, and introduce a temperature protection mechanism to prevent overheating damage; Furthermore, a pressure sensor is used to collect the actual pressure at the outlet of the plunger pump in real time; The flow sensor collects the actual flow rate through the system; Temperature sensors collect motor winding and hydraulic oil temperatures; Set the corresponding target pressure, target flow and safety temperature limit according to the upper control instructions or preset working conditions; Define two fuzzy input variables, pressure error and pressure error change rate, and define the fuzzy output variable PID parameter gain correction factor to construct a membership function set; Define the standard PID control law, the expression is: ; in, is the controller output, , , are the basic proportional, integral and differential coefficients respectively, is the current pressure error, is the historical error sequence; By introducing fuzzy logic output to perform weighted adjustment on PID parameters, an improved control law is obtained, which is expressed as: ; in, is the improved control law; The output of the improved PID controller Converted into a PWM waveform with adjustable duty cycle, used to control the motor inverter and thus adjust the motor speed; By changing the motor speed, the pressure output and flow characteristics of the plunger pump can be dynamically adjusted; Dynamic feedback regulation also includes the introduction of a temperature protection mechanism to achieve multi-dimensional coordinated control, including: Define the temperature anomaly flag function, the expression is: ; when When the temperature protection mechanism is triggered, the PWM duty cycle is forced to be reduced, which is expressed as follows: ; in, Temperature suppression coefficient, is the maximum tolerable temperature of the system, is the PWM duty cycle after temperature correction; It should be noted that the gain correction strategy using fuzzy logic to optimize PID control parameters enables the controller to dynamically adjust the response characteristics according to different operating states, thereby improving the regulation accuracy and system stability. At the same time, the introduced temperature protection mechanism can automatically reduce the motor output power under high-temperature conditions to avoid overheating and damage of the equipment, ensuring the safe operation of the system under continuous high-intensity work, and is applicable to complex working conditions such as high temperature and high pressure.

[0031] S5. Real-time display of the current working state on the monitor, providing a live view through the camera, automatically popping up a warning screen and playing a voice reminder when an emergency event is detected, and at the same time supporting remote alarm notification and log recording; Furthermore, define an event trigger function, and the expression is: ; Among them, is the event trigger flag. If any one of the sensor abnormality flags is activated, it is determined as an emergency event; When = 1, execute actions, and the actions include: popping up a red warning box on the main interface of the monitor, starting the voice synthesis module, playing a preset voice reminder statement, writing the sensor data and abnormal information at the current moment into the local memory, and sending an alarm signal to the management terminal through the wireless communication module; The voice reminder content is dynamically selected according to the abnormality type, and the expression is as follows: ; Among them, is the finally synthesized voice content, is the basic statement template, , , is the conditional splicing statement segment; It should be noted that through the event-driven alarm mechanism, combined with the camera image acquisition and voice synthesis module, the visual warning and voice reminder functions for emergency events are realized. The mechanism supports multiple feedback methods such as local pop-up prompts, voice broadcasts, log records, and remote alarm notifications, enhancing the intuitiveness and response speed of human-computer interaction, and is especially suitable for industrial field application scenarios that require rapid intervention and remote monitoring.

[0032] This embodiment also provides an electric pump control system based on instruction perception, including: A voice recognition module, a communication management module, a priority decision module, a status monitoring module, a dynamic regulation module, and a human-computer interaction module; The voice recognition module is used to start the microphone array and preprocess the environmental audio using an AI-driven noise reduction algorithm, and use an improved hidden Markov model to identify keywords in the audio signal and output a voice command signal; A communication management module, which is used to establish a two-way communication link with a remote controller, parse the content of control instructions, and dynamically adjust the transmission power, frequency hopping strategy or coding method according to the communication channel quality to ensure stable communication; A priority decision module, which is used to receive voice instruction signals and remote controller instruction signals, combine the pressure, flow rate, and temperature data collected by sensors, set multi-source instruction priority rules, and select and execute corresponding operations according to the priority; A status monitoring module, which is used to collect pressure, flow rate, and temperature parameters in real time, upload them to the main control unit after filtering, and use the LSTM machine learning model to predict trends and determine whether there are abnormal fluctuations; A dynamic regulation module, which is used to adopt a PID control algorithm combined with fuzzy logic to optimize the control curve according to the sensor feedback data, dynamically adjust the motor speed and the pump outlet pressure, realize closed-loop feedback regulation, and introduce a temperature protection mechanism to prevent overheating damage; A human-computer interaction module, which is used to display the current working status on a display in real time, provide a live view through a camera, automatically pop up a warning screen and play a voice reminder when an emergency is detected, and at the same time support remote alarm notification and log recording.

[0033] This embodiment also provides a computer device, which is applicable to the situation of an electric pump control method based on instruction perception, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the electric pump control method based on instruction perception proposed in the above embodiment.

[0034] This computer device can be a terminal. This computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication) or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball or a touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0035] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for controlling an electric pump based on instruction perception as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM for short), electrically erasable programmable read-only memory (EEPROM for short), erasable programmable read-only memory (EPROM for short), programmable read-only memory (PROM for short), read-only memory (ROM for short), magnetic memory, flash memory, a magnetic disk or an optical disc.

[0036] In summary, the present invention performs time-domain enhancement processing on the original audio by introducing an AI-driven noise reduction algorithm, effectively suppressing environmental noise and improving the clarity of the speech signal. By constructing an improved hidden Markov model that integrates an attention mechanism and a state transition weighting function, the semantic relevance between adjacent speech frames is enhanced, and the state transition path selection process is optimized, significantly improving the accuracy and robustness of keyword recognition. By constructing a priority scoring function that includes weight coefficients, dynamically evaluating the importance of different instructions in combination with sensor feedback data, and using a maximum value selection strategy to determine the final executed instruction, it is ensured that the system can give priority to responding to the operation with the highest safety and strongest rationality in any situation, realizing an automatic priority intervention mechanism in emergency situations.

[0037] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An electric pump control method based on instruction perception, characterized in that: Including: Start the voice recognition module, wireless communication module and sensor acquisition module, and establish a two-way communication link with the remote controller; Preprocess the ambient audio collected through the microphone array using an AI-driven noise reduction algorithm, and perform keyword recognition on the preprocessed audio using an improved hidden Markov model to obtain a voice command signal; The remote controller sends a control instruction to the receiving end, parses the instruction content and records the current communication channel quality, and dynamically adjusts the transmission power, frequency hopping strategy or coding method according to the channel quality to ensure stable communication between the remote controller and the electric pump; The main control unit simultaneously receives the voice command signal and the remote controller command signal, combines the data collected by the sensor, sets the instruction priority rule, and performs corresponding operations according to the priority; According to the instruction content, send a PWM signal to the motor controller to adjust the motor speed, and control the pressure output or flow rate of the plunger pump; Real-time collect pressure, flow rate and temperature parameters, upload the data to the main control unit after filtering, and use a machine learning model for trend prediction to judge whether there is abnormal fluctuation; The main control unit dynamically adjusts the motor speed and the pump outlet pressure according to the sensor feedback data, and uses the PID control algorithm combined with fuzzy logic to optimize the control curve to achieve dynamic feedback regulation; The display real-time displays the current working status and provides a live view through the camera. If an emergency occurs, a warning screen will automatically pop up and a voice reminder will be played.

2. The electric pump control method based on instruction perception according to claim 1, wherein: The improved hidden Markov model performs keyword recognition on the preprocessed audio to obtain a voice command signal. The specific steps are as follows: Perform time-domain enhancement processing on the received ambient audio signal to obtain a denoised pure audio signal; Perform short-time Fourier transform (STF) on the audio signal, extract the mel-frequency cepstral coefficients of each frame, and form an acoustic feature vector sequence; Define the improved hidden Markov model as a five-tuple, and the expression is: ； wherein, is the initial state distribution, is the state transition matrix, is the observation probability matrix, is the set of attention weights, is the state transition weighting function for dynamically adjusting the state transition probability; [[ID=IS=13]]Introduce attention weights to enhance the semantic connection between adjacent frames. The expression is: ; Among them, is the cosine similarity between the frame and the previous frame, is the attention scaling factor, is the attention weight of the frame, indicating the importance of this frame in the overall recognition process; Define the state transition weighting function, and the expression is: ; Among them, is the total number of hidden states in the model, is the state jump inhibition factor, represents the state and the state the distance between them; The state transition weighting function is used to punish cross-state jumps and encourage the model to recognize voice commands along a reasonable path; Multiply the state transition matrix in the original HMM by the weighting function to obtain a new state transition matrix; Based on the original observation probability, introduce attention weights Weight the observation values of each frame to update the observation probability matrix; Use the Viterbi algorithm to decode the improved hidden Markov model, find the hidden state sequence, and map it to the keyword recognition result; The expression of the Viterbi algorithm is: ; Among them, represents the maximum path probability at time in state . is the updated state transition probability, is the updated observation probability, is the observation value of the frame. The set of keywords recognized by the final output , including: start, stop, and speed regulation, and sending them as voice command signals to the main control unit.

3. The electric pump control method based on instruction perception according to claim 2, wherein: Combined with the data collected by the sensor, set the instruction priority rule, and perform corresponding operations according to the priority. The specific steps are as follows: Use a pressure sensor to monitor the pressure at the pump outlet end in real time; Use a flow sensor to measure the volume of liquid flowing through the pipeline; Use a temperature sensor to collect the working temperature of the motor and the hydraulic system; Perform denoising processing on the collected original data through a low-pass filter respectively to obtain a smoothed sensor signal sequence; Define an anomaly detection function to identify whether the current working condition is in a dangerous state. The expression is: ; wherein, is the maximum allowable pressure value, , are the minimum and maximum allowable flow rate ranges respectively, is the maximum allowable temperature of the motor or hydraulic system, , , are Boolean flags indicating whether the pressure, flow rate, and temperature exceed the safety thresholds respectively; Receive three types of control inputs, including: remote control instructions from the remote controller, voice instructions from the voice recognition module, and automatic control instructions generated by the main control unit according to sensor feedback; Define the instruction priority function, and the expression is: ; Among them, is the instruction priority score, indicates whether it is from the security system, indicates whether it is from the remote control, indicates whether it is from the voice command, > > , is the weight coefficient; At each moment , the main control unit simultaneously receives , , three instructions, and calculates their corresponding priority scores respectively , , ; Compare the magnitudes of the three and select the instruction corresponding to the largest one as the final execution instruction; Convert the selected final execution instruction into specific control signals, including: sending a PWM waveform to the motor controller to adjust the motor speed, controlling the solenoid valve to adjust the output pressure or flow rate of the plunger pump, and triggering the alarm device or the display to update the working status information.

4. The electric pump control method based on instruction perception according to claim 3, characterized in that: The pressure, flow rate, and temperature parameters are collected in real time, and the data is uploaded to the main control unit after filtering. A machine learning model is used for trend prediction to determine whether there are abnormal fluctuations. The specific steps are as follows: Use the smoothed sensor signal sequence as the input variable of the trend prediction model; Define the time window length centered on the current moment as, extract the historical information of the past N frames from the smoothed sensor signal sequence, and construct a multi-dimensional time series feature vector; Adopt the long short-term memory network LSTM as the trend prediction model. Its structure includes an input layer, a hidden layer, and an output layer. The input layer receives the feature vector, the hidden layer consists of multiple LSTM units, and the output layer generates the predicted value for the next moment. The expression is: ; Among them, represents the LSTM neural network mapping function, is the set of model parameters, including the weight matrix and bias terms, for the next time period is the predicted value; In the model training stage, use the historical sensor data set and minimize the mean square error loss function. The expression is: ; Among them, is the actual observed value, is the model predicted value, is the overall prediction error of the model; Define the residual function to represent the deviation between the current predicted value and the actual value. The expression is: ； Among them, is the actual sensor value at the current moment, is the predicted value of the model for the current moment, is the prediction error; Further introduce the moving average residual and standard deviation to adaptively adjust the anomaly detection threshold; For the sensor parameters, perform the above prediction and anomaly detection processes respectively, and output two types of information, including: trend prediction results and anomaly status flags.

5. The electric pump control method based on instruction perception according to claim 4, characterized in that: The main control unit dynamically adjusts the motor speed and the pump outlet pressure according to the sensor feedback data, and uses the PID control algorithm combined with fuzzy logic to optimize the control curve to achieve dynamic feedback regulation. The specific steps are as follows: Use a pressure sensor to collect the actual pressure at the outlet end of the plunger pump in real time; A flow sensor collects the actual flow rate flowing through the system; A temperature sensor collects the motor winding temperature and the hydraulic oil temperature; Set the corresponding target pressure, target flow rate, and safety temperature upper limit according to the upper-layer control instruction or the preset working condition; Define two fuzzy input variables of pressure error and pressure error change rate, and at the same time define the fuzzy output variable PID parameter gain correction factor to construct a membership function set; Define the standard PID control law. The expression is: ; Among them, is the output of the controller, , , are the basic proportional, integral, and differential coefficients respectively, is the current pressure error, is the historical error sequence; Introduce the fuzzy logic output to weight-adjust the PID parameters to obtain an improved control law. The expression is: ; Among them, is an improved control law; Convert the output of the improved PID controller into a PWM waveform with adjustable duty cycle for controlling the motor frequency converter and then adjusting the motor speed; Dynamically adjust the pressure output and flow characteristics of the plunger pump by changing the motor speed.

6. The instruction perception-based electric pump control method according to claim 5, wherein: In the dynamic feedback regulation, a temperature protection mechanism is also introduced to achieve multi-dimensional collaborative control, including: Define the temperature anomaly flag function. The expression is: ; When occurs, the temperature protection mechanism is triggered to force the PWM duty cycle to be lowered, expressed as follows: ; Among them, Temperature suppression coefficient, is the highest tolerable temperature of the system, is the PWM duty cycle after temperature correction.

7. The instruction perception-based electric pump control method according to claim 6, characterized in that: When an emergency event occurs, an alarm screen will pop up automatically and a voice reminder will be played. The specific steps are as follows: Define the event trigger function. The expression is: ； Among them, is an event trigger flag. If any sensor anomaly flag is activated, it is determined as an emergency event; When = 1, perform actions, where the performing actions include: pop up a red warning box on the main interface of the display, start the speech synthesis module, play a preset voice reminder statement, write the sensor data and abnormal information at the current moment into the local memory, and send an alarm signal to the management terminal through the wireless communication module; The content of the voice reminder is dynamically selected according to the anomaly type. The expression is as follows: ; Among them, is the final synthesized speech content, is the basic statement template, , , is the conditional splicing statement segment.

8. An electric pump control system based on instruction perception, based on the electric pump control method based on instruction perception according to any one of claims 1 to 7, characterized in that: Include: A voice recognition module, a communication management module, a priority decision module, a status monitoring module, a dynamic regulation module, and a human-machine interaction module; The voice recognition module is used to activate the microphone array, preprocess the ambient audio using an AI-driven noise reduction algorithm, identify keywords in the audio signal using an improved hidden Markov model, and output a voice command signal; The communication management module is used to establish a two-way communication link with the remote controller, parse the content of the control instruction, and dynamically adjust the transmission power, frequency hopping strategy, or coding method according to the communication channel quality to ensure stable communication; The priority decision module is used to receive the voice command signal and the remote controller command signal, combine the pressure, flow rate, and temperature data collected by the sensor, set the multi-source instruction priority rule, and select and execute the corresponding operation according to the priority; The status monitoring module is used to collect the pressure, flow rate, and temperature parameters in real time, upload them to the main control unit after filtering, and use the LSTM machine learning model for trend prediction to determine whether there is abnormal fluctuation; The dynamic regulation module is used to adjust the motor speed and the pump outlet pressure dynamically according to the sensor feedback data, adopt the PID control algorithm combined with fuzzy logic to optimize the control curve, realize closed-loop feedback regulation, and introduce a temperature protection mechanism to prevent overheating damage; The human-computer interaction module is used to display the current working status on the display in real time, provide a live view through the camera, automatically pop up a warning screen and play a voice reminder when an emergency is detected, and support remote alarm notification and logging at the same time.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the instruction-aware electric pump control method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the instruction-aware electric pump control method according to any one of claims 1 to 7.

Images