CN112908355B

CN112908355B - System and method for quantitatively evaluating teaching skills of teacher and teacher

Info

Publication number: CN112908355B
Application number: CN202110064599.5A
Authority: CN
Inventors: 闫长春; 彭长德; 魏明生; 王群
Original assignee: Jiangsu Normal University
Current assignee: Jiangsu Normal University
Priority date: 2021-01-18
Filing date: 2021-01-18
Publication date: 2022-02-25
Anticipated expiration: 2041-01-18
Also published as: CN112908355A

Abstract

The invention discloses a system and a method for quantitatively evaluating teaching skills of teachers and professors, belonging to the technical field of education and training and comprising the steps of constructing a teacher evaluation database; detecting the state of a teacher in real time through hardware; converting the teacher state into an evaluation data value; and comparing the evaluation data value with a teacher evaluation database for scoring, and comprehensively scoring and evaluating the teaching of the teacher to obtain an evaluation result of classroom performance of the teacher. The intelligent teacher-class training system can utilize big data processing to construct an objective evaluation system, an intelligent teacher and a real class are virtualized by combining artificial intelligence and modern sensing technology, and trainees can effectively solve the problems of high training time-space cost, insufficient teacher guide resources, lack of objective evaluation and the like and can also realize classified standard-reaching training, so that the purpose of individual accurate training is achieved, the training efficiency is effectively improved, and the quality of teacher-class training is further improved.

Description

System and method for quantitatively evaluating teaching skills of teacher and teacher

Technical Field

The invention relates to the technical field of education and training, in particular to a system and a method for quantitatively evaluating teaching skills of teachers and norms.

Background

The education quality can be improved after the development of education, and the education quality depends on the level of teachers. The level of the teacher affects the quality of a class. An excellent teacher is the key for successful teaching, and the teaching behavior of the teacher has important influence on the learning effect of students. Currently, most of the primary and secondary school teachers in China are from teaching in education. The professor as the reserve strength of the teacher team influences the development of future education. The education quality is improved, and the teaching method not only focuses on the construction of the teachers in the profession, but also focuses on the culture link of the education in the profession. Classroom teaching is a 'main battlefield' for teachers to teach knowledge and skills, and for teachers and students, how to teach the knowledge concepts, principles and viewpoints of a lesson by using clear languages, well-formed blackboard writing and good teaching gestures to guide the students to learn is the most basic skill for classroom teaching. On the basis of the teaching skill, other teaching skills can be further shown. As the operas train to sing, read, do and play, the three basic teaching skills of the teacher, namely the language, the blackboard writing and the teaching state, are the core elements for learning other teaching skills and are the 'basic work' training which is required before the teacher becomes a qualified teacher.

At the present stage, the teacher is mainly cultured in a teacher-guiding and student-training mode. The requirement of the training mode on teachers is high, so that teachers capable of guiding teachers to conduct teaching skill training are few, and the teachers are difficult to consider each teachers and the professors in the face of increasing numbers of the teachers and the professors. In basic skill training, students need to guide teachers to give real-time evaluations in order to clarify the direction of effort. It is a subjective qualitative assessment by teachers through personal feelings, such as: the sound is bright, the book is standard, the teaching state is natural and elegant, and the like. Such evaluation lacks clear objectivity standards, and no quantifiable objective evaluation system is formed, so students are often unsuitable, and need to try for a plurality of times after a long time to meet the requirements of teachers. Under the condition, the operability of the traditional method for implementing basic teaching skill training is poor, an effective training mode is lacked, large-scale effective training is difficult to implement, and the factors seriously influence the quality of teacher-teacher culture.

Disclosure of Invention

1. Technical problem to be solved

Aiming at the problems in the prior art, the invention aims to provide a teacher-model student simulation evaluation system and a teacher-model student simulation evaluation method, which can utilize big data processing to construct an objective evaluation system, combine artificial intelligence and modern sensing technology to simulate an intelligent teacher and a real classroom, and adopt an immersion type autonomous training method for trainees, so that the problems of high training time and space cost, insufficient teacher guide resources, lack of objective evaluation and the like are effectively solved, and meanwhile, graded standard-reaching training can be realized, the purpose of individual accurate training is achieved, the training efficiency is effectively improved, and the training quality of teacher-model student cultivation is further improved.

2. Technical scheme

In order to solve the above problems, the present invention adopts the following technical solutions.

A teacher teaching skill quantitative evaluation system comprises a cloud server, a virtual classroom, human-computer interaction equipment, a language detection module, a blackboard writing detection module and a teaching posture detection module, wherein the human-computer interaction equipment, the language detection module, the blackboard writing detection module and the teaching posture detection module are installed in the virtual classroom, the virtual classroom is composed of a high-performance engineering laser projector and a curved-surface large screen and is provided with self-adaptive 3D glasses, sound equipment and voice recognition equipment, the cloud server comprises a data acquisition module, a data processing module and a data comparison module, the human-computer interaction equipment is connected with the data comparison module through a network, the language detection module comprises voice processing equipment connected with the human-computer interaction equipment, the voice processing equipment is connected with voice recording equipment, the blackboard writing detection module comprises image processing equipment connected with the human-computer interaction equipment, the image processing equipment is connected with an intelligent conference flat plate, the teaching gesture detection module comprises action data processing equipment connected with human-computer interaction equipment, and the action data processing equipment is connected with a group of somatosensory inertial sensors.

A method for quantitatively evaluating teaching skills of teacher students comprises the following steps: the method comprises the following steps:

s1, constructing a teacher evaluation database;

s2, detecting the state of the teacher in real time through hardware; converting the teacher state into an evaluation data value;

and S3, the evaluation data value is compared with the teacher evaluation database for scoring, and the teaching of the teacher is comprehensively scored and evaluated to obtain the evaluation result of the classroom performance of the teacher.

The teacher evaluation database comprises an evaluation database of classroom language, an evaluation database of blackboard writing and an evaluation database of teaching gestures.

The professor teaching skill quantitative evaluation method according to claim 1, characterized in that: the evaluation data values comprise language detection data values, blackboard writing detection data values and gesture detection data values.

The professor teaching skill quantitative evaluation method as claimed in claim 1, characterized in that: the language detection data values comprise voice loudness, loudness variation degree, speech speed variation and whites. .

3. Advantageous effects

Compared with the prior art, the invention has the advantages that:

(1) according to the scheme, an objective evaluation system can be constructed by utilizing big data processing, artificial intelligence is combined, an intelligent teacher and a real classroom are virtualized by the modern sensing technology, and the trainees adopt an immersion type autonomous training method, so that the problems of high cost of training time and space, insufficient resource of a guide teacher and lack of objective evaluation and the like are effectively solved, and meanwhile, classified standard-reaching training can be realized, the purpose of individual accurate training is achieved, the training efficiency is effectively improved, and further the quality of teacher-model student cultivation is improved.

(2) The virtual classroom comprises high performance engineering laser projector and curved surface large screen, be equipped with self-adaptation 3D glasses simultaneously, sound equipment and speech recognition equipment, utilize projection and huge curved surface curtain, show whole picture in the front of the eye, form three-dimensional classroom picture through 3D technique and self-adaptation 3D glasses simultaneously, obtain classroom environment sound through sound equipment, carry out human-computer interaction through the speech recognition instruction, through two kinds of circumstances of self-adaptation lecture and writing board book of self-adaptation 3D glasses, the classroom environment that has the sense of immersion has been constructed, make the people obtain the sense of immersion and the sense of substitution of extremely coming, promote the teaching experience in the training process of model life.

(3) The motion data processing equipment is electrically connected with the intelligent conference tablet, and the motion of an arm can be more accurately judged to be the motion of scraping a blackboard or writing a blackboard through butting the intelligent conference tablet, and the basic method is as follows: when the action occurs, if a writing track is generated, the action is judged to be a writing action; when the action occurs, the writing track is cleared, and then the action of scraping the blackboard is judged. Meanwhile, the intelligent conference tablet is accessed, errors caused by irrelevant actions can be reduced, and the detection accuracy is improved.

(4) The voice recording equipment adopts a collar clamp type wireless microphone, so that teachers can conveniently move within a proper range in the teaching training process, the teaching immersion feeling is effectively improved, the voice processing equipment is internally connected with a voice noise reduction module, before the voice processing equipment processes and analyzes voice data recorded by the voice recording equipment, the voice data is subjected to noise reduction processing through the voice noise reduction module, and the accuracy of the voice processing equipment on a voice data processing result is effectively improved.

Drawings

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

FIG. 2 is a flow chart of the operation of the system of the present invention;

FIG. 3 is a basic teaching skill evaluation system profile of the present invention;

FIG. 4 is a distribution diagram of somatosensory inertial sensor positions of the present invention;

FIG. 5 is a map of the position of the head and neck sensors of the present invention;

FIG. 6 is a schematic view of the head and neck sensors of the present invention detecting a head-up and head-down;

FIG. 7 is a schematic view of the head and neck sensors of the present invention detecting face orientation;

FIG. 8 is a schematic structural diagram of adaptive 3D glasses according to the present invention;

FIG. 9 is a schematic structural diagram of adaptive 3D glasses according to the present invention;

fig. 10 is a schematic structural diagram of the adaptive 3D glasses according to the present invention when the lenses are opened.

The reference numbers in the figures illustrate:

the system comprises a cloud server 1, a data acquisition module 101, a data processing module 102, a data comparison module 103, a virtual classroom 2, a mirror frame 201, a accommodating box 202, a port 203, a lens 204, a support 205, a drive box 206, an electric push rod 207, a linkage block 208, a microcontroller 209, a human-computer interaction device 3, a language-4 detection module, a voice recording device 401, a voice processing device 402, a writing-on-board detection module 5, a intelligent conference tablet 501, an image processing device 502, a teaching posture detection module 6, a somatosensory inertial sensor 601 and an action data processing device 602.

Detailed Description

The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention; it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and all other embodiments obtained by those skilled in the art without any inventive work are within the scope of the present invention.

Example 1:

referring to fig. 1-7, a teacher teaching skill quantitative evaluation system comprises a cloud server 1, a virtual classroom 2, and a human-machine interaction device 3, a language detection module 4, a writing board detection module 5, and a teaching posture detection module 6 installed in the virtual classroom 2, wherein the virtual classroom 2 is composed of a high-performance engineering laser projector and a curved large screen, meanwhile, the device is provided with self-adaptive 3D glasses, sound equipment and voice recognition equipment, the whole picture is displayed in front of eyes by utilizing projection and a giant curved surface curtain, meanwhile, a three-dimensional classroom picture is formed by a 3D technology and self-adaptive 3D glasses, classroom environment sound is obtained by sound equipment, carry out human-computer interaction through speech recognition instruction, constructed the classroom environment that has the sense of immersing, made the people obtain the sense of extremely immersing and substitute the sense, promoted the teaching experience among the training process of professor student.

Referring to fig. 8-10, the adaptive 3D glasses include a frame 201, a pair of accommodating boxes 202 are fixedly connected to the front end of the frame 201, through holes 203 matched with the frame 201 are drilled in the outer ends of the accommodating boxes 202, lenses 204 matched with the accommodating boxes 202 are arranged in the accommodating boxes 202, the upper ends of the lenses 204 extend to the upper sides of the accommodating boxes 202, supports 205 are fixedly connected to the upper ends of the lenses 204, a pair of driving boxes 206 are fixedly connected to the front end of the accommodating boxes 202, the two driving boxes 206 are respectively located at the left and right sides of the through holes 203, electric push rods 207 are fixedly installed in the driving boxes 206, a pair of linkage blocks 208 located right above the driving boxes 206 are fixedly connected to the front ends of the supports 205, output shafts of the electric push rods 207 penetrate through the upper walls of the driving boxes 206 and are fixedly connected to linkage blocks 208 at corresponding positions, a microcontroller 209 electrically connected to an action data processing device 602 is fixedly installed at the outer ends of the frame 201, and the electric push rods 207 are electrically connected to the microcontroller 209, the microcontroller 209 and the motion data processing device 602 are connected through a wireless network or bluetooth, please refer to fig. 8, when the motion data processing device 602 detects that a teacher speaks to a student (i.e., facing a curved large screen), the microcontroller 209 controls the electric push rod 207 to be in a contraction state, so that the lens 204 is positioned in front of the teacher's eye (i.e., at the through port 203), so that the teacher's eye can obtain a three-dimensional classroom image, please refer to fig. 10, when the motion data processing device 602 detects that the teacher writes a blackboard writing (i.e., back to curved surface), the microcontroller 209 controls the electric push rod 207 to be in an extension state, so that the lens 204 leaves from the front of the teacher's eye, thereby effectively avoiding the lens 204 from interfering with the blackboard writing of the teacher's student, and realizing the adaptive adjustment function of the adaptive 3D glasses.

Referring to fig. 1-3, the cloud server 1 includes a data acquisition module 101, a data processing module 102, and a data comparison module 103, and the human-computer interaction device 3 is connected to the data comparison module 103 through a network. The data acquisition module 101 is used for acquiring video and audio data of domestic excellent teachers on class on the network, and the data processing module 102 is used for performing big data processing and analysis on the video and audio data, establishing a basic teaching skill evaluation system, and determining a teaching language skill quantification index, a teaching blackboard writing skill quantification index and a teaching posture skill quantification index. The teaching language skill quantification index comprises the following steps: speech rate, loudness, rate of change of speech rate, rate of change of loudness, speech whiteout, spoken language, speech standard (i.e., mandarin standard degree); the technical and energy-based indexes of the teaching blackboard writing comprise: word size, word spacing, line inclination, word definition degree and word specification degree; the teaching posture skill quantification index comprises the following steps: walking steps, distance and speed, turn-back times, head gestures (reflecting the times and amplitude of head raising and lowering of the teacher), facial orientations (reflecting the frequency and amplitude of left and right scanning of students by the teacher), and gesture actions (reflecting the frequency of hand actions of the teacher scraping a blackboard, writing a blackboard and the like).

Referring to fig. 1, the language detection module 4 includes a voice processing device 402 connected to the human-computer interaction device 3, the voice processing device 402 is connected to a voice recording device 401, the voice recording device 401 employs a collar-clip type wireless microphone, which facilitates teachers to move within a proper range during teaching and training, and effectively improves teaching immersion, the voice processing device 402 is internally connected with a voice noise reduction module, before the voice processing device 402 processes and analyzes voice data recorded by the voice recording device 401, the voice data is subjected to noise reduction processing by the voice noise reduction module, and accuracy of the voice data processing result by the voice processing device 402 is effectively improved.

Referring to fig. 1, the blackboard-writing detection module 5 includes an image processing device 502 connected to the human-computer interaction device 3, the image processing device 502 is connected to an intelligent conference tablet 501, a pressure sensor is installed inside the intelligent conference tablet 501, and an intelligent writing pen is provided, so that handwriting can be written immediately, a good writing feeling is achieved, and meanwhile, the track data of the blackboard-writing is recorded under the cooperation of the pressure sensor, so as to form blackboard-writing image data.

Referring to fig. 1, the teaching posture detection module 6 includes a motion data processing device 602 connected to the human-computer interaction device 3, the motion data processing device 602 is connected to a set of somatosensory inertial sensors 601, referring to fig. 4-5, the set of somatosensory inertial sensors 601 includes a head sensor, a neck sensor, a back sensor, two elbow sensors and two shoulder sensors, wearing positions of the head sensor and the neck sensor are shown in fig. 5, and the back sensor is worn on a back spine.

Referring to fig. 4-5, the back sensor is used to detect the number of foldback and walking steps, distance and speed, and for the number of foldback, the rotation of the back sensor is used to detect: when the back sensor rotates 180 degrees, the turning motion is recorded as one turning motion, and the turning motions of two times are recorded as one turning back. For the number of steps, distance and speed of walking, the swing of the back sensor along with the spine is detected: when a person walks, the spine swings left and right. When the left foot is stepped, the gravity center of the person moves towards the left front, and the spine also moves towards the left front; when the user steps forward, the center of gravity of the user moves to the right front, and the spine moves to the right front. According to the characteristics, the number of left-right swinging times is obtained through the back sensor, so that the walking step number can be obtained, and one step is counted when one swinging time is finished; obtaining walking pace through the period of the swing; and obtaining the walking distance through the displacement track of the back sensor. Referring to fig. 6-7, the head sensor and the neck sensor are used to detect the head pose and the face orientation, and referring to fig. 6, the head pose is detected by the rotation angle of the head sensor relative to the neck sensor on the vertical plane (i.e., the X-Z plane); referring to FIG. 7, for face orientation, the orientation is detected by the angle of rotation of the head sensor relative to the neck sensor in the horizontal plane (i.e., the X-Y plane).

Referring to fig. 4 to 5, the elbow sensor and the shoulder sensor are used to detect gesture actions, and whether writing or erasing is performed, the connecting line between the elbow sensor and the shoulder sensor is rotated by a larger angle than the initial position, and whether writing or erasing is performed can be distinguished by collecting the angle change. Referring to fig. 1 to 7, the motion data processing device 602 is electrically connected to the intelligent conference tablet 501, and by docking the intelligent conference tablet 501, it can more accurately determine whether the arm motion is a blackboard-scraping motion or a writing motion, and the basic method is as follows: when the action occurs, if a writing track is generated, the action is judged to be a writing action; when the action occurs, the writing track is cleared, and then the action of scraping the blackboard is judged. Meanwhile, the intelligent conference tablet 501 is accessed, so that errors caused by irrelevant actions can be reduced, and the detection accuracy is improved.

Referring to fig. 1-7, when a teacher needs teaching simulation training, first connect the voice processing device 402, the image processing device 502 and the motion data processing device 602 to the human-computer interaction device 3, and wear the voice recording device 401 and the seven somatosensory inertial sensors 601, where the seven somatosensory inertial sensors 601 are respectively worn on the head, elbows, shoulders and back of a user, and then the user needs to calibrate the seven somatosensory inertial sensors 601, and the specific calibration method is as follows: the user maintains the posture of standing upright and naturally hanging down the arm for three seconds, so that each somatosensory inertial sensor 601 establishes a coordinate system.

Referring to fig. 1-7, the teacher can then perform simulation teaching in the virtual classroom 2, so that the teacher can obtain extreme immersion and substitution, and the teaching experience in the training process of the teacher is improved. In the course of teaching simulation of a model student, the voice recording device 401 records teaching voice, and the voice processing device 402 processes and analyzes the recorded voice data to obtain values of speed, loudness, speed change rate, loudness change rate, voice margin, spoken language and voice standard in the course of teaching simulation of a model student; the intelligent conference tablet 501 records the teaching blackboard writing to form blackboard writing image data, and then the image processing equipment 502 processes and analyzes the blackboard writing image data to obtain numerical values of the character size, the character spacing, the line inclination, the character definition degree and the character specification degree of the blackboard writing; the motion data processing device 602 judges the motion made by the tester through the change of the self coordinate and the relative coordinate change of the seven somatosensory inertial sensors 601, and obtains numerical values of walking step number, distance and speed, turn-back times, head posture, face orientation and gesture motion in the teacher-student simulation teaching process.

Referring to fig. 1-7, after the teacher-model student simulation teaching is completed, the human-computer interaction device 3 can automatically input the language detection data, the blackboard-writing detection data and the teaching posture detection data into the data comparison module 103 of the cloud server 1, the data comparison module 103 can compare and perform deviation analysis on the input data and the quantization indexes to obtain evaluation levels of the teacher-model student teaching language skills, the teaching blackboard-writing skills and the teaching posture skills, and the evaluation levels are fed back to the teacher-model student through the human-computer interaction device 3. In conclusion, the system can utilize big data processing to construct an objective evaluation system, an intelligent teacher and a real classroom are virtualized by combining artificial intelligence and modern sensing technology, the participants adopt an immersion type autonomous training method, the problems of high training time and space cost, insufficient teacher guide resources, lack of objective evaluation and the like are effectively solved, meanwhile, graded standard-reaching training can be realized, the purpose of individual accurate training is achieved, the training efficiency is effectively improved, and the quality of teacher-model student culture is further improved.

The specific method for quantitatively evaluating the teaching skills of the teacher and the model students comprises the following steps: the method comprises the following steps:

s1, constructing a teacher evaluation database;

The evaluation data values comprise language detection data values, blackboard writing detection data values and gesture detection data values.

The language detection data values comprise voice loudness, loudness variation degree, speech speed variation and whites. Hardware and operating environment for voice detection

Software requirements: visual Studio 2013, Cool Edit Pro 2.1

Hardware requirements: pickup with single microphone attribute

Cool Edit Pro 2.1 is a software for converting sound signals into spectrogram, and the Visual Studio 2013 is used to write a program for analyzing the spectrogram so as to detect various indexes in classroom language.

The loudness of sound is an index for measuring sound energy when sound waves vibrate, has psychological perception characteristics related to the actual strength of sound signals, and is a measure provided for the strength of different sounds sensed by human ears, and the magnitude of the psychological perception characteristics is not only related to the amplitude of the audio signals, but also related to the frequency of the audio models. The loudness measurement of classroom language of the teacher is realized by adopting the loudness measurement algorithm principle of the audio signal in ITU-RBS1771 standard published by the radio and television bureau.

This method of loudness measurement comprises four phases: filtering, calculating the mean square energy of the period, calculating the loudness in the period, and calculating the average loudness in the test time.

The first step is to filter the electrical signal to reject waveforms due to environmental noise.

The second step is to determine the measurement period of the filtered signal

And calculating the audio mean energy value of the time block. Where the length of a time block is set

. During a measurement period

Inner to the first

The mean square energy value in each time block is shown in formula (1).

……………………………（1）

In the formula

Is the input signal after filtering.

The third step is to calculate the loudness of the audio data within the time block, where

The loudness calculation formula in each time block is shown as the following formula

……………………………（2）

The fourth step is to calculate the average loudness in the test time T

……………………………………（3）

Wherein

。

When a teacher gives a lecture, the sound should have fluctuation, namely the loudness changes, and the purposes of highlighting the teaching key point, drawing the attention of students and the like are achieved through the loudness changes. The variance of the loudness can be used for reflecting the loudness change, and the bigger the variance of the loudness is, the more obvious the change of the loudness is. It is known from the above section that loudness is essentially the mean square energy of sound waves. The variance of the mean squared energy per unit time is calculated to represent the change in loudness.

Herein the following

Is a unit time block, and has 50 small time blocks in a unit time

The first two steps are obtained by referring to the previous section to process the loudness of sound

The first in unit time

Mean square energy within small time block

The third step is to calculate

The mean square energy of each small time block in unit time,

the fourth step is to calculate the variance of the mean square energy of each unit time, firstly, several unit times and the number of unit times in the test time are calculated

(reduction of the decimal part)

The mean square energy per unit time averaged over the test time is then calculated as shown below

Finally, the mean square energy variance in the test time is calculated

Is the variance of the loudness we want to measure, i.e. the change in loudness.

The speed of speech is the vocabulary volume that people include per unit time when expressing or propagating information using vocabularies with propagation or communication meanings. Theoretically, the speech rate detection can be performed by recognizing the content of the speech of the tester, converting it into text information, and calculating the number of words appearing per unit time to obtain the speech rate value. Although the speech rate measured by the method is accurate, in the detection method, a conversion process from sound waves to texts exists, so that the detection time is increased, and the technical difficulty is increased.

After reviewing the literature, we found that each word in the chinese pronunciation is composed of a vowel phoneme and a consonant phoneme. In a spectrogram, a peak represents a vowel phoneme because the energy of a vowel phoneme is much greater than the energy of a consonant phoneme.

The speech rate to be measured can be obtained by calculating the number of wave peaks in unit time. We then convert the computation of speech rate to the computation of the number of peaks.

Since there may be pauses in the speech process, this part of the time should not be taken into account in the speech rate detection, otherwise the speech rate measurement will be seriously affected. We have then also considered this problem. By viewing a large amount of classroom video, it is statistically determined that when the teacher pauses for more than about 3 seconds, the portion of time is substantially the intentional pause time and should be recorded as invalid time.

The second step is to count the number of peaks within the time block. Since the energy at the peak is much greater than the energy at the trough, we can use thresholding to determine the absolute threshold above

The time block of (a) is counted. Set threshold value

The number of the counted time blocks is the number of the vocalization of the vowel phoneme and is recorded as

。

The third step is to calculate the effective time

In which time of invalidation

Wherein

The number of time blocks continuously exceeding 30 failed thresholds, namely the pause of exceeding 3s, and the time of the part is not counted into the time for calculating the speech rate.

The fourth step is to calculate the number of vowel phoneme per minute, which is the speed of speech to be measured.

The degree of the change of the speech speed reflects the change of the rhythm of the speaker, and the change of the speech speed in the class can be used for improving the attention of students. We often used mild or severe urgency to evaluate an excellent speech rate change. According to the method for processing loudness change in the foregoing, for the change of the speech speed, I choose to calculate the variance of the speech speed in the test time to represent the change size of the speech speed.

The speech rate in the ith time block is obtained by the method for calculating the speech rate in the first two steps according to the above

Wherein

Is at the first

The number of vowel phonemes occurring within a time block. And calculating the average speech rate in the test period

Thirdly, calculating the variance of the speech rate,

wherein

Refers to the number of time blocks that occur within the validity time of the test. Here shown at 10

Is a unit time block. Theoretically, the faster and more frequent the speech rate changes, the higher the value detected.

The voice margin is a pause part in a voice, and in the actual teaching, a teacher gives a certain pause to students for a certain thinking time. Therefore, a certain percentage of the white space is left in a speech stream. After observing a large number of classroom teaching videos and then analyzing the whites in the classroom videos, we record the time of pause more than 5s as whites. The first step is to filter the interference caused by ambient noise.

Setting a threshold, counting the number of time blocks lower than the threshold value f, and screening out the number of time blocks continuously exceeding 50 and not reaching the threshold

The time is the time left blank.

The third step is to calculate the time of the blank

。

The fourth step is to calculate the blank time to account for the test time

In a ratio of

。

Hardware introduction of blackboard writing detection:

the main body of the detection system is a MAXHUB intelligent large screen end, and the detection system is an intelligent writing terminal with pressure sensing, and screen sensing utilizes a capacitive touch screen principle. The intelligent writing pen is matched, strokes with different thicknesses can be displayed according to the force exertion degree of people during writing, writing information is converted into a writing track sequence, and then data processing is carried out locally.

The size of the characters on the blackboard writing is determined by the height and width of the characters,

writing on a writing screen in the first step, inputting an ordered track generated by writing into a writing board detection system, and stopping for 0.5s between writing two characters in order to ensure that a computer can cut two different characters.

And secondly, converting the acquired track information into a coordinate sequence, generating the coordinate sequence by one character, and judging that the writing of one character is finished and the generation of the coordinate sequence is finished if no writing track is generated for more than 0.5 s. Let the coordinate sequence of this word be

。

The third step is to calculate the width and height of the word, taken in this coordinate sequence

As the left boundary for this word,

to the right boundary of the word, the word is wide

Is calculated by the formula

Get again

As a lower boundary of this word,

for the upper boundary of the word, the word is high

Is calculated by the formula

The fourth step is to calculate the size of the character, which is the area of the rectangle enclosed by the upper, lower, left and right boundaries of the character, so that the size of the character

Is calculated by the formula

This is a calculation method for the size of a single word, and the size of processing many words is an average of the sizes of all words.

For the line spacing detection scheme, the first idea is to fit the lowest points of each character in the first line of characters into a straight line, and then fit the highest points of the second line of characters into a straight line. The average distance of the two straight lines is then calculated. However, in the actual measurement process, we find that if a word with a particularly long stroke or a word with a particularly short stroke exists in a segment of the text, the measurement result is greatly influenced. Then, we remove the coordinate value with larger dispersion in the two text segments. The resulting test values do not differ much from the actual values after reducing the dispersion.

After generating the coordinate sequence, obtaining the lowest point set of each word in the first line as

The highest point set of each word of the second line is

. In a line of text, either

Or also

The number of elements in (1) is not large, so that generally, only one minimum value and one maximum value need to be removed to achieve the purpose of reducing the dispersion.

After the dispersion is reduced, two straight lines are obtained by fitting a curve,

the first straight line is:

the second straight line is:

let the left boundary of the longest row be

The right boundary is

。

So the average distance of the two straight lines

Is shown as the formula

The word spacing reflects the size of the interval between two written words, and when the interval is too small, the two words can be squeezed mutually; too large a space also results in too large a space between words, which affects the aesthetic feeling. Detection of word spacing and writing on blackboardThe wide detection method is similar, and the blackboard writing word spacing is determined by the right boundary of the previous word and the left boundary of the next word. However, there is a difference from the detection of the word size, that is, when the word space between two lines of words is processed, if the head and the tail of the two lines of words are not distinguished, the error of the word space measured by calculation is large. Therefore, the line head and the line tail are distinguished in the detection, and the distance between the line head and the line tail is not calculated. The detection scheme comprises the following steps: the large screen end collects writing information to generate two coordinate sequences,

Then, take the right boundary of the first word, i.e.

Then take the left boundary of the second word

So that the two words are spaced apart by

。

In the same way we can get the second word

And the third word

The word spacing of

. Until the m-th word is calculated, when the m-th word and the n-th word (

) If the distance of (a) is less than a (a is a negative constant) or greater than b (b is a positive constant), it is determined that the writing of the word in the line is completed, and the nth word is the head of the word in another line, so the word distance between the nth word and the word in another line is not calculated.

And finally, calculating the average value of all word intervals to obtain the average word interval.

When the line spacing is calculated, the line spacing is calculated by calculating the average distance of two fitting curves, and the line inclination can be obtained by calculating the slope of the fitting curves. Let the slope of the fitting curve be k, then the line slope

. Since the average row inclination is calculated for each row, all calculated row inclinations are first calculated

And taking absolute values, and then carrying out summation and averaging. When calculating the line inclination, the same problem as calculating the line spacing is that the too long stroke and the too short stroke affect the result, so we also perform the de-dispersion process on the lowest point of each word to obtain the fitting curve.

For the detection of the recognition degree of writing, the first idea is to convert the writing track into an internal code and then compare the internal code with a standard internal code library of Chinese characters by using a network. However, since the number of characters is large compared to the number of english alphabets, it takes a long time to detect the characters, and it takes about one minute to recognize several characters.

Therefore, the detection method is improved, namely a fixed text given by the system is written before detection, so that the system only needs to compare the internal code generated by the writing track of the text with the standard internal code. The detection time is greatly accelerated. Meanwhile, through the step, the writing habits of the tester can be collected, such as the time required for writing one word, the pause time between two words, the relative length of each stroke and the like. This also helps in the recognition of formal writing.

When the same pen is used, the writing definition is the visual embodiment of the writing strength, so that the writing definition can be obtained only by measuring the writing strength. The collection equipment of writing dynamics is the pen that writes that has pressure sensing, and the inside of pen has piezoresistor, when writing with the dynamics of difference, and the inside electric current of pen can change, and the signal that transmits to blackboard writing system terminal through the bluetooth is also different, shows the thickness degree of telecommunication through converting into the stroke on intelligent large-size screen again at last to after the test, the definition degree of exporting every word.

The motion of human body is mainly caught by the sensor to the detection of religion appearance to signal conduction computer terminal with the sensor collection carries out data processing through the bluetooth.

The seven sensors are all three-axis acceleration sensors. Respectively worn on the head, the elbow, the shoulder and the back of a tester,

before each sensor is started to detect, the position of each sensor needs to be corrected. During correction, the tester needs to stand upright and maintain the posture of the arm hanging down naturally for three seconds. At this point, each sensor will establish a respective coordinate system.

In the front of standing

The positive direction of the axis and the right-hand direction are

The positive direction of the axis, vertically upwards, is

Positive direction of the axis. The sensor judges the action made by the tester according to the change of the coordinate of the sensor and the relative change of the coordinate.

The method for detecting the walking steps is a plurality of methods, most commonly, the walking steps are detected in mobile phone software, the method is to count the steps by using a gravity accelerometer in a mobile phone to sense the direction and the size of the gravity change, but the method has low detection accuracy and is difficult to test the speed of each walking step. The swing of the spine is mainly used herein in the detection of the number of steps. The sensor used is an acceleration sensor at the back spine. When a person walks, the spine swings left and right. When stepping left foot, the center of gravity of the personThe spine can move towards the left front; when the user steps forward, the center of gravity of the user moves to the right front, and the spine moves to the right front. According to the characteristic, the walking steps can be obtained only by acquiring the left-right swinging times through the sensor, and one swinging step is counted. By the period of the swing, we can also get the walking pace

. Setting the period of the measured oscillation to

Then the step speed is

。

The detection of the walking distance is also completed by a sensor on the back, zero calibration is carried out before testing, and the default zero calibration place of the sensor is used as the coordinate origin

. When the walking-type robot walks in the same direction,

the value of the shaft basically does not change greatly, and the sensor is mainly used in

Move on a parallel plane, and set the coordinates of walking to any point as

We need only to calculate

The distance moved can be obtained. In this case, the distance measurement is inaccurate, considering that it is impossible to move only in one direction during actual movement. We then combine the measured steps to arrive at a distance detection schemeThe improvement is made.

When detecting the total walking distance, the walking distance of a single step is firstly measured, and the coordinate of a sensor after walking the step is set as

So that the walking distance of one step is

(ii) a After the second step of walking, the coordinates of the sensor are

Then the walking distance of the second step is

；

Similarly, the moving distance of the nth step is

If only the measured step number of each step is added, the total step number of the walking is obtained

。

The speed is detected in two terms, one is walking pace and the other is walking speed. The walking pace has been described in the foregoing step number detection method. For the speed of the ambulation we obtain it by calculating the ratio of the ambulation distance to the time of the spinal swing. The walking time is not the total time of the test, since the person is not always in a state of walking when actually used. It is reasonable to use the time of the spinal swing instead of the walking time. Suppose the time of the spinal swing is

Then speed of walking

The calculation formula is as follows:

，

the head posture detection is used for detecting the frequency and amplitude of head raising and head lowering and is mainly realized by a head sensor and a neck sensor together. Since the head raising and lowering movements are the swinging movements of the head about a point on the neck, these movements result in a change in the relative positions of the two sensors.

The position of the neck sensor is set as the origin coordinate, and when the head sensor is kept in the head-up front, no matter where the head sensor walks, the head sensor is kept still relative to the neck sensor,

the coordinates of the head sensor at this time are

Connection of two sensors with

The axis forming an angle of

(ii) a The coordinates of the head sensor will change when the head is lowered, and the coordinates are

Two sensors are connected to

The angle of the axis being

The angle of rotation is

The angle of the sensor is

The angle of rotation is

The angle of rotation is positive when raising the head. When the frequency of head raising and head lowering is calculated, in order to eliminate the influence of slight displacement of a sensor on the calculation frequency, the head raising times are only counted when the rotation amplitude exceeds 5 degrees; for the rotating amplitude exceeding-3

The number of head-down times is counted. The detection of the face orientation is mainly used for reflecting the frequency and the amplitude of teachers and students sweeping across the eyes, and is realized by calculating the frequency of head swing and the amplitude of left-right swing and by means of a head sensor and a neck sensor. Similar to the detection of the head-up and head-down frequency and amplitude, when the orientation of the face of the subject changes, the relative positions of the two sensors also change.

Let the initial position of the head sensor be

When the orientation of the face is changed, that is, the head is swung to the right or left, the coordinates of the head sensor become

(ii) a When swinging to the left, the coordinates of the head sensor become

. When the face swings left and right, the value of the z coordinate of the head sensor does not change, and the angle rotated by the projection of the connecting line of the two sensors on the x-y plane can be used for calculating the face rotation amplitude. So that the right pendulum has a turning amplitude of

The amplitude of the right swing is

. When rotating overWhen the amplitude is larger than 10 degrees, the right head swing or the left head swing is recorded once, and the positive sign and the negative sign of the calculated amplitude are used for judging whether the head swing is right head swing or left head swing.

The times of scraping the blackboard and writing are collected by an elbow sensor and a shoulder sensor, and no matter writing on the blackboard or scraping the blackboard, a connecting line of the elbow and the shoulder sensor rotates by a larger angle compared with an initial position, so that whether actions of writing on the blackboard and scraping the blackboard are carried out can be distinguished by collecting the angle change of the connecting line, and therefore a scheme is designed to collect the angle change of the connecting line;

the zero calibration shoulder sensor is used as a zero coordinate, and the position of the sensor at the elbow is set as

When the sensor at the elbow is moved during the blackboard erasing or writing, the coordinate of the elbow sensor is

But the relative distance from the shoulder sensor is always kept as

. A distance from the position of the time correction unit itself is

Then, we can use the cosine theorem to calculate the rotated angle, which is

。

And counting the times of scraping the blackboard and writing the blackboard once as long as the angle is larger than a certain value phi, and calculating the time of keeping the rotation angle larger than phi to record as the time of scraping the blackboard or writing the blackboard once. The flow chart of the detection is as follows

By simulating the actions of scraping the blackboard and writing the blackboard for multiple times, phi is set to be 30 degrees, and the detection accuracy is higher by taking 30 degrees as a limit for distinguishing the actions of scraping the blackboard and writing the blackboard from the irrelevant actions.

At present, the single gesture teaching system cannot distinguish actions of scraping the blackboard and writing, and can judge whether the action of lifting the arm by 30 degrees is the action of scraping the blackboard or writing the blackboard by butting the blackboard writing system. The basic method comprises the following steps: when the action occurs, if a writing track is generated, the action is judged to be a writing action; if the writing trace is cleared while the motion is occurring, it is determined as a blackboard-scraping motion. Meanwhile, the board writing system can reduce errors caused by irrelevant actions and increase the detection accuracy.

In a classroom, proper turning back is helpful for teaching to keep vitality and enhancing the interaction with students. The fold return motion is acquired by a sensor on the back. When the sensor rotates 180 degrees, the turning motion is recorded as one turning motion, and the turning motions of two times are recorded as one turning back.

Simulation intelligent detection system construction principle

Principle of scientificity

The development of the artificial intelligence evaluation system needs the support of a computer technology and a network technology, the evaluation process of the artificial intelligence technology needs to rely on big data, the system continuously simulates thinking modes and behaviors of people, and the artificial intelligence evaluation technology can be completed only when the big data reaches a certain amount. The artificial intelligence evaluation is a very complex technology, the evaluation process contains many contents, and the artificial intelligence evaluation has common parts with multiple sciences, wherein other disciplines such as linguistics, psychology and computer disciplines are included, and the intelligent embodiment needs support of multiple technologies. Therefore, the artificial intelligence evaluation is used, namely, the teaching skill is evaluated more scientifically and objectively by artificial intelligence, and more accurate, more convincing and more objective evaluation is completed by replacing people with real data analysis.

Principle of objectivity

One of the advantages of the intelligent detection and evaluation system is that the cost of school human resources is saved, and efficient evaluation is realized. With the continuous improvement of modern science and technology, the basic teaching evaluation mode gradually becomes diversified, and a plurality of modes which are difficult to evaluate in the past can be solved through an intelligent evaluation system, such as the speed of speech and the size of sound. Generally, the sense of the user is taken as an evaluation standard, but the degree of the bottom speech speed is more or less standard, the loudness of the sound is most appropriate, and the user cannot objectively speak. However, the voice information can be completely mastered by collecting big data through the artificial intelligence evaluation system, utilizing a corresponding operation model and adopting an artificial intelligence technology, and meanwhile, the data can be continuously updated by utilizing network management, so that further information guarantee is realized.

The artificial intelligence evaluation system is not invariable, has the ability of continuous learning, can continuously put a lot of new data into the system, processes the data with low operation difficulty, and then obtains data with higher levels. Due to the continuous learning function of artificial intelligence, the level of network management is continuously improved, the running speed is also continuously accelerated, and meanwhile, the time of an evaluation process is reduced.

Principle of expert type

The modern society is a big data era, the evaluation system is gradually intelligentized, and the intelligent evaluation system establishes a comprehensive evaluation system by utilizing the solving technology of a computer through an expert knowledge base in a computer network. The characteristics of different voices, characters and action information output by people can be realized, and the evaluation quality and efficiency can be improved only by accelerating the intelligent process of the evaluation system. The expert system is an important component of modern artificial intelligence, and can bring knowledge information and experience provided by experts, scholars and teachers in excellent classes into the artificial intelligence evaluation system, then the artificial intelligence technology processes the information, and finally the information is utilized to evaluate the voice, blackboard writing and teaching states of the tested person.

Principle of high efficiency

By applying the expert system, the efficiency of computer network management can be improved. The artificial intelligence evaluation system network version can realize cloud evaluation, is not limited in schools and classrooms, and can realize teacher and model student teaching skill evaluation in dormitories, playgrounds and homes. And the artificial intelligence evaluation system network version takes the internal knowledge base as a basis for analyzing and processing data, and then rapidly completes the management task. The artificial intelligence evaluation system is a technology for automatically searching data information after finishing inputting voice, teaching state and writing information. The system will automatically transmit the searched data to a specific location to provide an intelligent service. For example, after a teacher who performs teaching evaluation inputs his own voice information, the artificial intelligence technology can apply corresponding search information to process, and then analyze data according to the information, and display the evaluation result or provide information required by the teacher, so as to save the time for evaluating teaching skills.

Principle of applicability

The artificial intelligence evaluation system has been widely applied in the current life, for example, in daily life and work, the applied automatic mail collection, online shopping, meeting arrangement, route planning and the like all use the artificial intelligence evaluation function, a big data system for each user is formed through the collection of big data, the big data is analyzed and evaluated according to the intelligent evaluation system, and an optimal presentation scheme is selected. Meanwhile, the artificial intelligence evaluation system is applied to a plurality of enterprises at present, and convenience is brought to enterprise management. The artificial intelligence evaluation system can realize automatic management, so that the management of enterprises is more informationized, and the working efficiency of enterprise personnel is improved. In the conventional evaluation mode, a large amount of expenditure is required, and the actual evaluation effect is not good, and the variation is likely to occur. By applying the artificial intelligence evaluation system, the cost budget problem of evaluation can be basically solved, and the enterprise management effect is improved. The artificial intelligence evaluation system can accumulate knowledge and experience of different professions and different industries, analyze and summarize the contents, then form a set of complete and scientific computer network intelligence evaluation system, improve the working efficiency, and therefore the artificial intelligence evaluation system has strong applicability.

An artificial intelligence evaluation system is a novel thing which is developed continuously by scientific technology and is in line with the era, is built and developed continuously by means of communication technology, computer technology and network technology, and the development is also applied to different fields gradually. In the future, artificial intelligence evaluation systems are also continuously updated, developed and perfected. The artificial intelligence evaluation system can better meet different requirements of people on work and life, the network technology can be continuously developed, the running speed of the network can be gradually accelerated, the safety degree can be continuously enhanced, the fields related to artificial intelligence can be continuously increased, and more values are created for social development

The teacher basic teaching skill quantitative evaluation system fully focuses on student attention by utilizing various training modes of simulation training and circular training in a quantitative evaluation form, evaluates student training results in a flexible and various mode, presents training samples with scientificity and proper difficulty in a gradual mode in the aspect of teacher and model student skill training, and properly guides students to complete systematic training; in the aspect of skill reinforcement, the traditional teacher-model student skill training cannot be reinforced in time by using various reinforcement modes, but a quantitative evaluation system of the basic teaching skill of a teacher can repeatedly train a certain skill of students in a targeted manner; in the aspect of training indexes, the definition of the traditional teacher and model student skill training indexes is not clear enough, and the traditional teacher and model student skill training indexes cannot accurately correspond to teaching skill training targets of teacher and model students. A new course reform improves, promotes the requirement to teacher's basic teaching skill training more, and as the modern teacher, future people's teacher should learn more consciously, train oneself teacher's basic teaching skill. The scientific evaluation basis is the premise of all capability improvement, and artificial intelligence education is the development trend of education modernization. The system takes the quantitative measurement and evaluation of the basic skills in the schoolmates classroom as a reference, analyzes the practical characteristics of the basic skill indexes of the schoolmates, and constructs a quantitative measurement and evaluation system according to the characteristics. The quantitative measurement and evaluation system is based on teaching voice skills, writing skills and gesture teaching skills, a refined multi-level question bank is used as an evaluation basis, quantized structural and non-structural learning records are used as data sources, and subdivided system evaluation is used as an evaluation result. The teacher can be guided by practice of the measured standard value of the key points of the basic teaching skills, and a self-aided and intelligent basic classroom skill training mode for the teacher can be provided. The system is expected to standardize the basic teaching skills of the teacher and the professor, help the teacher and the professor to adjust the teaching strategy according to the overall evaluation result, complement the short board according to the system evaluation result and realize the self value.

Quantitative evaluation is an indispensable important component in basic teaching skill training. Due to the limitation of human factors, the expected effect is difficult to achieve through qualitative evaluation, and the feasibility of training can be effectively improved through introducing quantitative research into experiments. Quantitative evaluation as one of the methods of empirical study, is substantially equivalent to the study of empirical significance, as indicated by sali khiskinson: "the demographer thinks the world is there, which is in a static form for researchers to study. Therefore, quantitative studies suggest that the material and social reality are independent of the observer himself, and it is possible to construct scientific knowledge as a result of observation and evaluation of the fact without bias. The voice of the student is intercepted and led into an experimental system, the evaluation of the system can be used as a sample, numerical values are constructed on the basis in an effort mode, other various video data are compared, an experimental result is obtained by using a system analysis method as far as possible, the similarity of each video is obtained to the maximum extent, a unified result is obtained, and the unified result is used as a standard value of a teacher-and-model language system.

Virtual simulation system teacher skill evaluation result

As can be seen by comparing the training results of the traditional skill training and teacher basic teaching skill quantitative evaluation system, the traditional training and teacher basic teaching skill quantitative evaluation system training has certain difference in teaching skill training of teacher students. In the aspect of blackboard writing skills, a traditional teacher and professor cannot evaluate the size, the specification, the depth, the line spacing, the inclination and the like of a word in a quantitative mode, the content of traditional evaluation is subjective, individual emotional factors are easy to participate, meanwhile, different people have different opinions on different training result evaluations, and the score of an index cannot be measured according to a specific standard. Meanwhile, a large amount of manpower and material resources are consumed in traditional evaluation, for example, group evaluation requires a plurality of people to be together at the same time, if the training time of one student is 15 minutes, a 4-person group needs at least one hour to finish listening, and then time is spent for evaluation and statistics. Also, expert evaluation is faced with such a situation, and there is a large difference between the number of experts and the number of students, it is impossible for each expert to give his own opinion to the students, and one-to-one guidance is not achievable in the case of the current shortage of educational resources.

Claims

1. A method for quantitatively evaluating teaching skills of teacher students comprises the following steps: the method comprises the following steps:

s1, constructing a teacher evaluation database,

s2, detecting the state of the teacher in real time through hardware; the teacher status is converted into an evaluation data value,

s3, comparing the evaluation data value with a teacher evaluation database for scoring, and comprehensively scoring and evaluating the teaching of the teacher to obtain the evaluation result of the classroom performance of the teacher;

the teacher evaluation database comprises an evaluation database of classroom language, an evaluation database of blackboard writing and an evaluation database of teaching gestures;

the evaluation data values comprise language detection data values, blackboard writing detection data values and gesture detection data values;

the language detection data values comprise voice loudness, loudness change degree, speech speed change and whiteout;

detection of change in loudness: here by T_d5s is a unit time block, and 50 small time blocks T exist in one unit time_g

The first step is to process the loudness of sound to obtain the mean square energy in the ith small time block in the jth unit time

In the formula (ii)_iIs a signal that is input after being filtered,

the second step is to calculate the average value of the mean square energy of each small time block in the jth unit time,

the third step is to calculate the variance of the mean square energy of each unit time, firstly, several unit times and the number of unit times in the test time are calculated

Finally, the mean square energy variance in the test time is calculated

S, the variance of the loudness to be measured, namely the change of the loudness;

the speech rate detection method comprises the following steps:

the first step is to filter the electric signal and filter out the waveform caused by the noise of the environment;

the second step is to count the number of wave crests in the time block, because the energy at the wave crest is much greater than the energy at the wave trough, the time block higher than the absolute threshold gamma is counted by using threshold processing, and the set threshold gamma is_InitialThe number of time blocks counted is the number of uttered vowel sounds, denoted as n, for 55LKFS；

The third step is to calculate the effective time T_{Is effective}＝T_{General assembly}-T_InvalidationIn which time of invalidation

Wherein q is the number of time blocks which continuously exceed 30 failing thresholds, namely a pause which exceeds 3s, and the time of the part is not counted into the time for calculating the speech rate;

the fourth step is to calculate the number of vowel phoneme per minute, which is the speed of speech to be measured

Detecting the variation of the speech rate, adopting the variance of the speech rate in the test time to express the variation of the speech rate, and obtaining the speech rate in the ith time block in the first two steps according to the method for calculating the speech rate

Wherein n is_iFor the number of vowel phonemes occurring within the ith time block, and calculating the average speech rate over the test period

Thirdly, calculating the variance of the speech rate,

wherein m refers to the number of time blocks appearing in the effective time of the test, and 10s is taken as a unit time block;

the blackboard writing detection data values comprise: word size, word spacing, line inclination, word definition degree, and word specification degree.