CN116146436A - Multi-mode monitoring system and method for fan operation - Google Patents

Abstract

The invention discloses a fan operation multi-mode monitoring system and a method, which belong to the technical field of wind power generation detection, wherein a plurality of information acquisition modules comprise an odor signal acquisition unit, a sound signal acquisition unit and an image signal acquisition unit, an information analysis processing module comprises an odor signal processing unit, a sound signal processing unit and an image signal processing unit, and a data comparison unit, wherein the odor signal processing unit, the sound signal processing unit and the image signal processing unit are used for receiving and processing information sent by the plurality of information acquisition modules and sending the processed information to the data comparison unit, the data comparison unit is used for receiving information sent by the odor signal processing unit, the sound signal processing unit and the image signal processing unit, comparing and analyzing the information, and an alarm module connected with the information analysis processing module is used for receiving the comparison and analysis results of the data comparison unit and sending an alarm, so that the problems of fault monitoring and real-time early warning in the fan operation are solved.

Description

Multi-mode monitoring system and method for fan operation

Technical Field

The invention belongs to the technical field of wind power generation detection, and particularly relates to a fan operation multi-mode monitoring system and method.

Background

Each fan is huge and complex in structure, the cost in various aspects is very high, the safety requirement is very severe, otherwise, the consequences are serious, so that fault monitoring and real-time early warning in the operation of the fan are important in the safety production of the fan, the types of the fan faults are various, the reasons are also very complex, the types and the working conditions of the fans are different, the faults are different, but the following faults are generally summarized: (1) The machine generates serious friction, vibration impact and other abnormal sounds, the foundation screw is broken, and the bedplate generates cracks; (2) lubricating oil overflows, goes bad or has burnt smell, and smokes; smoke or burnt smell of the motor, tripping of a power switch and the like; (3) The fan blade freezes when the temperature is low, and secondly, in the daily operation in-process of fan, fan tower section of thick bamboo also has following problem: (1) entering a tower by external personnel; (2) the operation and maintenance personnel safety protection equipment is not regularly worn; (3) In the process of maintenance and repair in the cabin of the operation and maintenance personnel, a skylight is privately opened and is discharged out of the cabin; (4) The maintainer opens the escape hole by mistake, so that the personnel falls off at high positions; (5) Before an maintainer enters the fan hub in the engine room to operate, a mechanical lock is not locked; (6) When the gearbox is inspected by an maintainer, goggles are not worn when the gearbox is opened, and the goggles are burnt by hot steam of the gearbox.

Disclosure of Invention

To solve the problems set forth in the background art. The invention provides a system and a method for monitoring fan operation in a multi-mode manner, which have the characteristics of multi-mode monitoring, edge calculation and real-time analysis and early warning of an AI deep learning algorithm.

In order to achieve the above purpose, the present invention provides the following technical solutions: a fan operation multi-mode monitoring system comprises

The information acquisition modules are arranged on the fan and comprise an odor signal acquisition unit, a sound signal acquisition unit and an image signal acquisition unit;

the information analysis processing module is connected with the information acquisition modules and comprises an odor signal processing unit, a sound signal processing unit and an image signal processing unit, and a data comparison unit, wherein the odor signal processing unit, the sound signal processing unit and the image signal processing unit are used for receiving and processing information sent by the information acquisition modules and sending the processed information to the data comparison unit, and the data comparison unit is used for receiving and comparing the information sent by the odor signal processing unit, the sound signal processing unit and the image signal processing unit;

and the alarm module is connected with the information analysis processing module and comprises an alarm center management platform and is used for receiving the comparison analysis result of the data comparison unit and giving an alarm.

Preferably, the information acquisition module is composed of a camera integrated with an odor sensor and a sound sensor, and the image signal acquisition unit is connected with a visual identification AIoT box and is installed in a plurality of areas of a fan operation site.

Preferably, the odor signal processing unit comprises an odor signal classification network and a numerical modeling, the sound signal processing system comprises waveform processing, anomaly detection and filtering, and the image signal processing unit comprises a heterogeneous network learning model and an unsafe behavior intelligent analysis system.

Preferably, the data comparison unit comprises a priori knowledge and a fuzzy neural network inference network.

Preferably, the information acquisition module and the information analysis processing module are connected through a 5G network or an Ethernet.

A method for multi-modal monitoring of fan operation, comprising:

the information acquisition module monitors the running site state and the environment of the fan in real time and sends real-time monitoring data to the information analysis processing module;

the information analysis processing module analyzes and compares the real-time monitoring data and sends normal or abnormal signals to the alarm module;

the alarm module displays the compared result and pushes abnormal information to the equipment, the mobile phone and the computer for alarm.

Compared with the prior art, the invention has the beneficial effects that:

1. the fan operation multi-mode monitoring system is installed in a plurality of areas of a fan operation site through integrating an odor sensor and a sound sensor by a camera, meanwhile, the camera is connected with a visual identification AIoT box, an integrated embedded industrial control main board and a neural network acceleration chip are mounted on the camera, and common interfaces such as Ethernet, USB and GPIO are integrated. The intelligent analysis task of monitoring videos is completed by loading a trained deep learning model (the training model needs to be carried out at a deep learning workstation) in application, faces are detected and identified (faces are automatically detected from monitoring pictures and face identification is carried out), smoke detection is carried out (whether fire alarms exist in a fan cabin or the like are detected), blades are detected (whether damage or icing exists in the fan cabin or the like is detected), helmets are detected (whether the helmets are worn by pedestrians or not is automatically detected), clothing is detected (working clothes, production of safety personal protection accessories or the like), emotion recognition (micro-expression analysis, emotion states of people are identified), intrusion detection (whether pedestrians break into forbidden areas, climbing up of the fence or the like is detected, electronic fences, virtual walls or the like are supported), falling behavior detection (whether pedestrian events occur or not is detected), behavior analysis (identification operation specifications, specific actions (e.g. throwing waste, object behavior analysis and the like), abnormal events (such as collisions, falling, fighting, falling, small group scattering, falling, group scattering) and the like) are carried out, and the intrusion amount is calculated, so that the intrusion amount is calculated, and the intrusion amount is compared with the detection module.

2. The fan operation multi-mode monitoring system described by the invention collects sound information through a plurality of information collecting modules, and then calculates effective values, peak factors, peak-to-effect ratio, kurtosis and other parameters of sound signals in a time domain through an information analysis processing module. By observing the spectra of the different samples in the frequency domain, some typical differences in spectral distribution of normal and faulty core important device signals are mined. And finally, carrying out nonlinear mapping on sample data by utilizing an artificial neural network, and solving the problem of space division of characteristic parameters, thereby classifying normal and abnormal information from sound signals and monitoring abnormal sound problems in the running process of the fan.

3. According to the fan operation multi-mode monitoring system, the image, sound and smell information are collected independently through the information collecting modules, and the information analyzing and processing module actively reports the health condition of equipment through real-time intelligent analysis, so that the equipment overhaul efficiency and effect are improved.

Drawings

FIG. 1 is a schematic diagram of a system framework of the present invention;

FIG. 2 is a schematic diagram of a voice detection network according to the present invention;

FIG. 3 is a frequency domain diagram of a sound waveform according to the present invention;

description of the embodiments

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Referring to fig. 1-3, the present invention provides the following technical solutions: a fan operation multi-mode monitoring system comprises

The information acquisition modules are arranged on the fan and comprise an odor signal acquisition unit, a sound signal acquisition unit and an image signal acquisition unit;

the information analysis processing module is connected with the information acquisition modules and comprises an odor signal processing unit, a sound signal processing unit and an image signal processing unit, and a data comparison unit, wherein the odor signal processing unit, the sound signal processing unit and the image signal processing unit are used for receiving and processing information sent by the information acquisition modules and sending the processed information to the data comparison unit, and the data comparison unit is used for receiving and comparing the information sent by the odor signal processing unit, the sound signal processing unit and the image signal processing unit;

and the alarm module is connected with the information analysis processing module and comprises an alarm center management platform and is used for receiving the comparison analysis result of the data comparison unit and giving an alarm.

Specifically, the information acquisition module is composed of a camera integrated smell sensor and a sound sensor, and the image signal acquisition unit is connected with a visual identification AIoT box and is installed in a plurality of areas of a fan operation site.

Specifically, the odor signal processing unit comprises an odor signal classification network and a numerical modeling, the sound signal processing system comprises waveform processing, anomaly detection and filtering, and the image signal processing unit comprises a heterogeneous network learning model and an unsafe behavior intelligent analysis system.

Specifically, the data comparison unit comprises priori knowledge and a fuzzy neural network reasoning network.

Specifically, the information acquisition module is connected with the information analysis processing module through a 5G network or an Ethernet.

A method for multi-modal monitoring of fan operation, comprising:

the information acquisition module monitors the running site state and the environment of the fan in real time and sends real-time monitoring data to the information analysis processing module;

the information analysis processing module analyzes and compares the real-time monitoring data and sends normal or abnormal signals to the alarm module;

the alarm module displays the compared result and pushes abnormal information to the equipment, the mobile phone and the computer for alarm.

Wherein the sound is collected

The analogic input () function in the Matlab function library can implement acquisition of analog signals, such as sound signals, voltage, current signals, etc. By defining ai=analog input ('ADAPTOR'), an object of analog signal acquisition can be established, wherein adaptor= 'wireless', i.e., an object of sound signal acquisition is established.

And setting attribute parameters for sound collection of the established sound signal collection object. ADDchannel (ai, 1) sets the number of channels in which sound is collected, 1 representing mono and [ 12 ] representing stereo. Set (ai, sampling rate', value) sets the sampling frequency of sound signal acquisition, the value of value is determined according to practical situations, but the sampling theorem must be satisfied, and the typical values are value=8000, 16000, 44100 and the like; set (ai, 'samplealert', value) sets the length of the collected sound signal; set (ai, 'trigger type', 'value') sets a triggering mode of sound signal acquisition, and the triggering modes mainly include three modes: the software triggering can be divided into signal rising edge triggering, falling edge triggering and the like, and an appropriate triggering mode is required to be selected according to different actual conditions.

The execution of the status (ai) function opens the object of sound signal acquisition, and once the triggering condition is met, the sound signal is acquired immediately, and the function getdata (ai) can continuously acquire the sound signal and the characteristic value, and the delete (ai) function deletes the object of sound signal acquisition. The collection of sound can be realized by executing the functions at one time

Program for sound collection:

ai=analoginput('winsound');

addchannel(ai,1);

ai.samplerate=8000;

ai.samplespertrigger=30000;

ai.triggertype='immediate';

start(ai);

[data,time] = getdata(ai);

plot(data);

title('shiyu');

xlabel('time');

ylabel('data');

grid on；

2. spectral analysis of fan operation sound signals

And carrying out FFT analysis on the time domain signal, wherein the FFT is a fast algorithm of fast Fourier transform, and is obtained by improving the discrete Fourier transform according to the characteristic of the discrete Fourier transform.

The functions FFT and IFFT are used for fast fourier transform and inverse transform in the matlab's signal processing toolbox. The function FFT is used for fast fourier transform of the sequence, where the format for the points is y=fft (x), where x is a vector, y is the FFT of x and is the same length as x; if x is a matrix, y is FFT for each column vector of the matrix.

Program for spectral analysis of sound signals:

F=fft(data);

f1=angle(data);

N=length(time);

w=[1:N].*5/N;

plot(w,abs(F));

3. filtering of sound signals

The filter is a device for eliminating interference noise, and the input or output is filtered to obtain pure alternating current. You can deduce the most general filter type from the basic filter building block-the second order general filter transfer function: low pass, band pass, high pass, band reject and elliptic filters. The design adopts a Butterworth IIR filter to filter the sound signals.

Wherein: wp represents the passband cut-off frequency; ws represents the stop band cut-off frequency, rp represents the passband ripple coefficient; rs represents the stop band ripple coefficient; n represents the filter minimum order; wn represents the cutoff frequency. b, a respectively represents a numerator and a denominator polynomial coefficient vector of a digital filter system transfer function with the order of n+1; fs is the sampling frequency; n is the number of frequency points selected in the frequency range of the section [ OFs ]; f, recording the frequency points. n to the power of 2 can increase the speed of the operation because the freqz function uses the FFT algorithm of basis 2. ftype=high, a high pass filter; ftype=band pass, a band pass filter; ftype=stop, a band reject filter.

4. The overall procedure is:

ai=analoginput('winsound');

addchannel(ai,1);

ai.samplerate=8000;

ai.samplespertrigger=30000;

ai.triggertype='immediate';

start(ai);

[data,time] = getdata(ai);

figure;

plot(data);

title('shiyu');

xlabel('time');

ylabel('data');

grid on

F=fft(data);

f1=angle(data);

N=length(time);

w=[1:N].*5/N;

figure;

plot(w,abs(F));

fp=1200%

Fs=1100%passband cut-off frequency

Ft=8000%

As=20;

Ap=1;

Wp=2*pi*Fp/Ft;

Ws=2*pi*Fs/Ft;

Fp=2*Ft*tan(Wp/2);

Fs=2*Ft*tan(Ws/2);

[n,wn]=buttord(Wp,Ws,Ap,As,'s');

[b,a]=butter(n,wn,'s');

[num,den]=bilinear(b,a,1);

[h,w]=freqz(num,den);

bu=filter(b,a,F);

figure;

plot(time,bu);

grid;

xlabel ('frequency');

yabel ('frequency response amplitude');

title ('filter');

waveform after operation: see fig. 3;

the collected fan operation sound is regarded as a special signal, namely a 'complex vector'. The fan operation sound is used as a voice signal stored in a computer and is a discretized vector, and the discretized vector is extracted and processed. This is achieved by using a powerful tool MATLAB for processing digital signals. A channel is established between the collected abnormal sound and the digital signal processing through the calling of a plurality of command functions in MATLAB;

1) Hardware description of sound collection system

The hardware system is mainly applied to industrial production, and the operation state and the like of fan equipment are detected by comparing the collected sound signals with data in a database. The system is mainly divided into the following parts: level conversion circuit, AD conversion circuit, static storage and dynamic storage, USB interface and JTAG part.

The hardware system detects cracks, adhesion and the like of fan blade and other parts by collecting sound signals. The capability of the DSP for processing the digital signals at high speed and the capability of the USB for transmitting data at high speed are combined, so that the DSP is used for service of fan operation and maintenance work, and the DSP is the main work content of the hardware system. The system considers that TMS320VC5402 of TI company is selected as the CPU of the PCB, and PDIUSBD12 of Philips company is used as an interface chip, and the USB1.1 protocol is used for communication between the DSP and the computer.

2) Hardware design scheme of sound collection system

The sound signal that can be heard by humans is an analog signal in the range of 20-20kHz, so that the sensor is first required to receive the sound signal, and then conversion is required to change the sound signal from an analog signal to a digital signal. And then detecting the covered sound signal by analyzing the cause and law of noise generation and utilizing the characteristics and coherence of the detected signal. The detection method comprises the methods of coherent detection of frequency domain signals, accumulation and average of time domain signals, counting technology of discrete signals, parallel detection and the like.

Because ROM and DRAM resources in 5402 sheets are limited, an external storage device is needed by the sound collection system, and the system selects an SRAM as a static memory and a FLASH as a dynamic storage device. 5402 has a CPU voltage of 3.3V and a peripheral voltage of 1.8V, so the system also requires a power supply module for supplying power, which can convert a general input voltage of 5V into a voltage of 3.3 and 1.8V to supply power to the DSP, and the 5V voltage can also supply power to other devices except the DSP.

The sound signal acquisition system based on TMS320VC5502 adopts a modified Harvard structure, and comprises 12 groups of independent buses, namely a1 group of program read buses, a1 group of program address buses, a 3 group of data read buses, a 2 group of data write buses and a 5 group of data address buses. This architecture allows simultaneous execution of program instructions and data operations, with fast execution speeds, single cycle fixed point instruction execution times of 10ns. In acoustic signal acquisition, the accuracy and real-time nature of the analog-to-digital signal conversion (ADC) plays an important role in subsequent signal processing. The TLC320AD50 is adopted in the design to complete the A/D conversion of the sound signal. The TLC320AD50 is a 32bit synchronous serial port A/D and D/A conversion chip provided by TI company, 1 extraction filter is arranged behind the ADC to improve the signal to noise ratio of the input signal, the highest sampling frequency can reach 22.5Kb/s, and the requirements of sound signal processing on the sampling frequency are met.

The DSP is communicated with the computer in a USB mode, the USB interface chip is directly connected with the DSP, and the USB protocol is realized through the program of the DSP.

In summary, in the system, several basic links are: level conversion circuit: converting a 5V power supply into 3.3V and 1.8V which respectively supply power to on-chip peripheral equipment of the DSP chip and the CPU; AD signal conversion circuit: converting an analog signal received by a sensor into a digital signal for processing by a DSP; a signal storage circuit: storing signals processed by the DSP; and a signal transmission circuit: uploading the processed signal to a server;

3) Introduction to modules

3.1 DSP

Introduction to DSP technology

The digital signal processor, DSP for short, is a chip for professional signal processing and has wide application in the communication and automatic control fields at present. Today, where information resources are greatly enriched, the degree of digitization is becoming higher and higher. DSP, as an important component of this technology, has had an increasing impact on our lives. Since the 1978 AMI company issued "single processing devices", the trend in today's DSP has been toward mixed architectures, with the distinction between DSP products and computers becoming increasingly ambiguous, from the first generation of general-purpose DSPs based on Harvard architecture but using different data and program buses, to the second generation of enhanced general-purpose DSPs that have been improved, to the third generation of DSPs that incorporate GPP architecture. In the background of the digital age, DSP has become a basic device in various fields such as electronic products, and its application in the fields of motor control, voice recognition and image recognition is more widespread.

DSP used in sound collection system

The DSP in the sound collecting hardware system adopts TMS320VC5402 (5402) of TI company, the operation speed is up to 100 MIPS, and the sound collecting hardware system has improved Harvard structure, so that the sound collecting hardware system can complete multiplication of 32x32bit in one instruction period, and can also complete multiplication and addition operation most commonly used in mathematical operation rapidly. It has 4 address buses, 3 16-bit data memory buses and 1 program memory bus, 40-bit arithmetic logic unit (AIU), a 17 x 17 multiplier and a 40-bit special adder. 8 auxiliary registers and a software stack, a C language compiler allowing the use of the most advanced fixed-point DSP, a built-in programmable wait state generator, a phase-locked loop (PLL) clock generator, two multi-channel buffer serial ports, an 8-bit parallel HPI port for communication with an external processor, 2 16-bit timers and a 6-channel DMA controller, and is particularly suitable for battery powered equipment.

3.2 level converting circuit

The level conversion circuit, as the name implies, converts the voltage supplied by the power supply into a voltage suitable for the chip to operate. Because 5402's nuclear voltage is different from on-chip peripheral voltage, and the voltage that whole circuit needs can not be provided by the power directly, so level conversion circuit can be said to be the power of whole circuit work, provides the condition that is fit for its work for each components and parts.

In this circuit, TPS767D301 (hereinafter referred to as D301) of TI company is used as the power supply chip. D301 is a chip that can output different voltages separately, and can output 3.3V and a certain regulated voltage between 1.5-5.5V. Since the peripheral voltage of 5402 is 3.3V and the core voltage is 1.8V, in this design, the output of the chip is set to 3.3V and 1.8V, matching 5402.

In the output portion of 1OUT vo= "Vref" × (1+r1/R2), vredf= 1.1834V in D301, so vo= "1".1834v× (1+15.8/30.1) =1.8V.

3.3 AD conversion

The AD conversion chip selected in the design is TLC320AD50C of TI company. The sampling of the chip uses ΣΔ technique, i.e. a sampling filter is placed after the ADC and a difference filter is placed before the DAC. The biggest characteristic of the structure is that the system can simultaneously perform receiving and transmitting tasks. The TLC320AD50C can realize AD/DA conversion with high sampling rate (up to 22.5 kb/s), and the function is realized by 2 synchronous serial conversion channels with 16 bits, and can directly communicate with a DSP connection.

The selectable items and circuit configurations in TLC320AD50C may be programmed through a serial port through which the chip may be programmed and configured for power down, reset, signal sampling rate, serial clock rate, gain control, communication protocol, test mode, etc.

The off-chip reset circuit provides power-on reset, and the crystal oscillator circuit can provide a main clock frequency of 10MHz, and the data sampling frequency and other clock signals are distributed by the frequency. The communication format between 5402 and AD50C is the primary serial communication format: receiving and transmitting the converted signal.

3.4 Storage of

After the sound signal is collected, an important link is the storage of the sound signal, and in the system, the FLASH memory of the SST company is adopted: SST39VF400A. The storage capacity of the device is configured according to actual collection, 3.3V single power supplies are adopted for supplying power, and the read-write and erasure of each submodule can be realized through a plurality of special command word sequences without providing high voltage additionally. In this design we use DSP programming to implement read and write operations to the memory.

The DSP accesses off-chip memory primarily through an External Memory Interface (EMIF). It has not only very strong interface capability (can interface with various memories directly) but also very high data throughput capability. 5402 and SST39VF400 are shown in fig. 1. The circuit controls the erasing, reading and writing of FLASH mainly through the relevant output pins of the DSP. Wherein A0-A19 are address lines, DQ 0-DQ 15 are data lines, OE and WE are output enable and write enable respectively, and CE1 is chip enable.

The sound signal is transmitted to the DSP after passing through the AD converter, the enabling ends of flash and sram are respectively controlled by PS and DS pins of the DSP through logic switches, and the RW and MSTRB control bits of the DSP are respectively used for controlling reading and writing through logic circuits.

In this scheme, the SRAM uses GS1117:64K x 16 1MB asynchronous static random access memory. GS71116 is a static random access memory consisting of high-speed complementary metal oxide semiconductor transistors (CMOS), requiring no external clock or time strobes. An operating voltage of 3.3V, all inputs and outputs are compatible with transistor logic circuits (TTL). Its fast channel time is less than 15ns and the operating current is less than 100mA.

3.5 USB

The PDIUSBD12 is a USB interface device with a parallel bus, which conforms to the USB specification of the 1.1 version of the universal serial bus, integrates SIE, FIFO, memory transceiver, voltage regulator, and the like, can realize a high-speed parallel interface of 2 mbytes/sec with any external microcontroller or microprocessor, can realize a data transmission rate of 1 mbyte/sec in both batch mode and synchronous mode, can control connection with the USB through software, adopts a connection indicator of GoodLink technology, blinks an LED during communication, has programmable clock frequency output, is internally powered on reset and low voltage reset circuit, is dual-power operation, can be used in 3.3±0.3V or an extended 5V power supply, and can realize batch and synchronous transmission in a multi-interrupt mode.

3.6 JTAG

JTAG is a short for joint test action group, which is an emulation part for debugging a DSP, and its connection part is consistent with pins on an emulator. JTAG pins are reserved specially for the DSP5000 series of TI company, 14 pins are grounded, the 4,8,10,12 pins are suspended, the 5-connection high-level voltage is 3.3V, all simulation pins use IEEE1149.1 standard, and the rest pins have the meanings of [ 5 ]: 1. TMS: an input pin for selecting a test mode; 2. TRST: an input pin for testing reset; 3. TDI: an input pin for inputting test data; 7. TDO: the output pin outputs data when the TCK is in a falling edge, and the rest time is in a high-resistance state; 9. tck_ret: the input pin is used for connecting the board with the simulator when the connection cable is not smaller than 6 inches, the connection method is the same as the TCK, and the input pin is required to be additionally driven when the connection method is larger than 6 inches; 11. TCK: an input pin, a test clock, typically an intrinsic clock signal with a duty cycle of 50%; 13. EMU0: the simulation interrupt pin 0 can be used as input or output; 14. EMU1: the emulation interrupt pin 1 can be used as an input or an output, and when TRST is low and EMU0 is high, EMU1 is low and all outputs are disabled.

4) Conclusion(s)

By the sound collection system, intangible sound signals are converted into graphs to be processed, and waveform characteristics of the intangible sound signals can be observed to be analyzed.

2. Processing software and hardware system for abnormal sound data of fan operation

The running state of the fan can be monitored in real time through abnormal sound data processing.

The Matlab language is computer application software with powerful data analysis and processing functions, and can convert sound file into discrete data file, process data with powerful matrix operation capacity, such as digital filtering, fourier transform, time domain and frequency domain analysis, sound playback, etc. and the signal processing and analysis tool box provides sound signal analysis with rich functions.

The system utilizes the basic principle of digital signalology to realize the processing of abnormal sound signals of the fan operation, and comprehensively utilizes the technologies of signal extraction, amplitude-frequency conversion, fourier transformation, filtering and the like to process the sound signals in the matlab7.0 environment.

Anti-electromagnetic interference design of acquisition equipment:

the sound sensor is small in size, high in integration level and dense in packaging, and is easy to be interfered by various kinds. The sources of interference are divided into: system internal and system external. The internal interference mainly comprises mutual interference of different power supply voltages, mutual interference among signals, interference of the power supply voltages on the signals and the like; external interference organic onboard secondary power supply, complete machine cable network, electric equipment interference in working environment and the like. The system performs electromagnetic interference processing in the following basic manner.

1) Photoelectric isolation is a typical signal isolation technology, and can avoid mutual interference of alternating current and direct current, strong current and weak current, and the like. The use of opto-electrical isolation for DIO channels in the system, such that a significant portion of the interference is blocked;

2) The filtering has two modes of inductive filtering and capacitive filtering, and can effectively inhibit the conductive interference from the public power supply. The system is provided with a filtering design between a power supply and the ground, shaping and filtering of various signals and the like on a bottom plate and an inserting plate of the controller;

3) When the equipment shell is designed, a high-reliability connector is selected for fastening installation. In addition, in order to effectively inhibit external radiation interference, a shielding shell is adopted, and a high-performance shielding wire is adopted for the transmission cable.

Data transmission

4.5.1 data transmission mainly comprises transmission of monitoring video and preprocessing results, and the transmission mode comprises wired optical fiber and wireless 5G network transmission. The performance targets of the 5G network are high data rate, delay reduction, energy saving, cost reduction, system capacity improvement and large-scale equipment connection, so that the monitoring video can be stably and rapidly transmitted to the back-end server by virtue of the advantages of the 5G technology, rapid analysis and processing are realized, response is made, and the availability and reliability of an unsafe behavior detection and early warning command platform are improved.

(1) Data analysis and processing

This section includes traditional monitor platforms that can use commercially mature software products such as: the system comprises an iVMS platform for Haikang vision, a monitoring software platform for bloom, a monitoring platform for Yu-View iMOS platform family, a monitoring platform for Darcy KDM series and the like. The SDK provided by the corresponding platform can be integrated with the security generation management and early warning system.

The intelligent analysis algorithm server is an artificial intelligent system based on GPU high-performance calculation, a distributed system and deep learning, and can conduct real-time intelligent analysis from the monitoring video, so that intelligent visual calculation tasks of automatically conducting unsafe behavior detection and analysis on a large-scale monitoring camera in real time are achieved. The platform can be integrated with a safety generation management and early warning system and other management systems through an open API, so that linkage of the whole system is realized.

(2) Results display and management system integration

The part mainly processes intelligent analysis results, such as result display, triggering alarm, equipment locking and the like. Meanwhile, the system can be integrated with other safety generation management systems and command platforms to process the business flow of the production process.

Wherein the collection of odors

When smoke generated in a fire place enters the monitoring ionization chamber during smoke sensing of the sensor, the detection source americium 241 in the ionization chamber emits a rays, so that air in the ionization chamber is separated into positive and negative ions. When smoke enters, the inner ionization chamber and the outer ionization chamber are opposite in polarity, so that the generated ion current is kept relatively stable and is in an equilibrium state; aerosol submicron particles and visible smoke released at the initial stage of fire disaster enter a detection ionization chamber in large quantity, adsorb and neutralize positive and negative ions, enable ionization current to be reduced sharply, change ionization balance state and output detection electric signals, and after being processed and identified by a later-stage circuit, alarm is sent out and an alarm switch signal is output to a matched monitoring system.

The procedure was as follows: # include < reg51.H ]

#define uchar unsigned char

sbit s1=P1^0;

sbit d1=P1^1;

sbit beep=P2^3;

void delay (uchar x)

{

uchar a,b;

for(a=x;a>0;a--)

for(b=100;b>0;b--);

}

void main()

{

s1=1;

d1=1;

delay(100);

while(1)

{

if(s1==1)

d1=0,beep=0;

else

d1=1,beep=1;

}

}

Single-limit comparator

The threshold voltage is changed by adjusting the slide rheostat, when the input voltage is changed (namely, when the sensor alarm works), different voltages are output, the P1-1 and the working state of the buzzer are controlled through P1-0-!

2. Audible and visual alarm circuit

Under the control of a 51 single chip microcomputer, when a P1-0 port inputs low level, the P1-1 and the buzzer are in a low level state and do not work; when the P1-0 port inputs high level, the P1-1 and buzzer are at high level, and work normally, namely alarm-!

By adopting the technical scheme, the fan operation multi-mode monitoring system and the method are characterized in that the information acquisition module is used for monitoring the fan operation field state and environment in real time and sending real-time monitoring data to the information analysis processing module;

the information analysis processing module analyzes and compares the real-time monitoring data and sends normal or abnormal signals to the alarm module;

the alarm module displays the compared result and pushes abnormal information to equipment, a mobile phone and a computer for alarm;

the system achieves the purposes that the image, sound and smell information are autonomously collected through the information collecting modules, and the information analyzing and processing module actively reports the health condition of the equipment through real-time intelligent analysis, so that the equipment overhaul efficiency and effect are improved.

Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (6)

1. A fan operation multi-mode monitoring system, characterized in that: comprising

The information acquisition modules are arranged on the fan and comprise an odor signal acquisition unit, a sound signal acquisition unit and an image signal acquisition unit;

the information analysis processing module is connected with the information acquisition modules and comprises an odor signal processing unit, a sound signal processing unit and an image signal processing unit, and a data comparison unit, wherein the odor signal processing unit, the sound signal processing unit and the image signal processing unit are used for receiving and processing information sent by the information acquisition modules and sending the processed information to the data comparison unit, and the data comparison unit is used for receiving and comparing the information sent by the odor signal processing unit, the sound signal processing unit and the image signal processing unit;

and the alarm module is connected with the information analysis processing module and comprises an alarm center management platform and is used for receiving the comparison analysis result of the data comparison unit and giving an alarm.

2. The blower operation multi-mode monitoring system of claim 1, wherein: the information acquisition module is formed by integrating an odor sensor and a sound sensor through a camera, and the image signal acquisition unit is connected with a visual identification AIoT box and is installed in a plurality of areas of a fan operation site.

3. The blower operation multi-mode monitoring system of claim 1, wherein: the smell signal processing unit comprises a smell signal classification network and a numerical modeling, the sound signal processing system comprises waveform processing, anomaly detection and filtering processing, and the image signal processing unit comprises a heterogeneous network learning model and an unsafe behavior intelligent analysis system.

4. The blower operation multi-mode monitoring system of claim 1, wherein: the data comparison unit comprises priori knowledge and a fuzzy neural network reasoning network.

5. The blower operation multi-mode monitoring system of claim 1, wherein: the information acquisition module is connected with the information analysis processing module through a 5G network or an Ethernet.

6. A method for multi-modal monitoring of fan operation, comprising:

the information acquisition module monitors the running site state and the environment of the fan in real time and sends real-time monitoring data to the information analysis processing module;

the information analysis processing module analyzes and compares the real-time monitoring data and sends normal or abnormal signals to the alarm module;

the alarm module displays the compared result and pushes abnormal information to the equipment, the mobile phone and the computer for alarm.

Images